id stringlengths 7 26 | text_type stringclasses 2
values | original_text stringlengths 25 7.75k | cot_response stringclasses 1
value | domain stringclasses 31
values | grpo_v5_3 stringlengths 310 815 | grpo_v7_1 stringlengths 371 841 | grpo_v7_2 stringlengths 389 844 | grpo_v7_3 stringlengths 342 872 | prompt listlengths 2 2 | ground_truth stringclasses 2
values | prompt_len int64 157 1.89k |
|---|---|---|---|---|---|---|---|---|---|---|---|
human-1443 | human | We are all capable of the horrendous act that Hitler committed or fostered in his life. The right circumstances and societal influences could bring out the same in each of us. I am definitely not supporting any of the atrocities he engaged in. But there is historical perspective that could perhaps explain some of his a... | NA | cmv | {
"rationale": "First, the passage shows a personal, inconsistent voice with contractions, self-deprecation, and hedged, tentative phrasing (e.g., 'I am definitely not...') typical of human online posts. Second, it contains small stylistic idiosyncrasies and repetition ('Again... again') and emotionally nuanced ambiv... | {
"rationale": "First, the passage shows a personal, conversational tone with hedging language ('I am definitely not...') and emotional nuance that align with human online discussion. Second, it contains small stylistic inconsistencies and colloquialisms ('But war is war', repeated 'again') suggesting spontaneous hum... | {
"rationale": "First, the passage shows a personal, conversational tone with hedging language ('I am definitely not...') and emotional nuance that align with human online discussion. Second, it contains small stylistic inconsistencies and colloquialisms ('But war is war', repeated 'again') suggesting spontaneous hum... | {
"rationale": "First, the passage displays a personal, emotionally nuanced voice with contractions and self-reflection (e.g., 'I am definitely not supporting...') that aligns with human online discussion. Second, it contains minor stylistic inconsistencies, colloquialisms ('Again, I am thoroughly disgusted...'), and... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 294 |
human-2139 | human | First of all, I do have a problem with the manner in which drone strikes are carried out by the US, mainly the broad scope and language the Authorization for Use of Military Force grants POTUS. I am not arguing for or against the drone program or the the lack of transparency in which individuals are targeted. What I am... | NA | cmv | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, argumentative tone typical of human online posts. Second, it shows human-like errors and idiosyncrasies — a duplicated word 'the the', inconsistent capitalization, and personal rhetorica... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, argumentative Reddit style consistent with human posting. Second, it shows typographical errors ('the the'), inconsistent spacing, and emotionally charged, loosely structured prose that ... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, argumentative Reddit tone consistent with human posting. Second, it shows typographical errors and repetition ('the the') and uneven phrasing suggesting casual human typing rather than p... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, anecdotal tone typical of human online posts. Second, it shows human-like errors and idiosyncrasies — a duplicated word ('the the'), casual phrasing ('where' for 'were'), and abrupt topi... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 540 |
human-1833 | human | I think that there is a definite double standard between men and women. I think that it's possible for women to objectify guys like Ryan Gosling and Channing Tatum, while, if a guy says that Carmen Electra is hot, he's a sexist pig. I think that women can hit men all they want, but if a man even pretends to hit a woman... | NA | cmv | {
"rationale": "First, the post uses a highly conversational, forum-specific voice (CMV, TL;DR) and includes community conventions that are typical of human social media posts. Second, it shows idiosyncratic humor, personal frustration, and culturally specific references (Ryan Gosling, Channing Tatum, Carmen Electra)... | {
"rationale": "First, the post uses a highly conversational, forum-specific voice (CMV, TL;DR) and includes community conventions that are typical of human social media posts. Second, it shows personal bias, emotional engagement, and idiosyncratic phrasings ('I'm also tired of...') suggesting spontaneous human expre... | {
"rationale": "First, the post uses highly conversational, community-specific markers (CMV, TL;DR) and a personal voice typical of Reddit-style human posts. Second, it shows idiosyncratic humor, emotional engagement, and uneven phrasing ('I'm also tired of...') suggesting spontaneous human expression rather than pol... | {
"rationale": "First, the post uses highly conversational, community-specific markers (CMV, TL;DR) and a personal voice suggesting a Reddit-style human poster. Second, it contains idiosyncratic phrasings, emotional tone, and culturally specific references (Ryan Gosling, Channing Tatum, Carmen Electra) consistent wit... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 326 |
human-1424 | human | "Blackface," a white male painting his face black to perform as an racist caricature of a black person goes back a few hundred years in one way or another but didn't really catch on until the 1830s with comic Thomas Rice's song "Jump Jim Crow." The song was accompanied by a dance with Rice in blackface as the character... | NA | cmv | {
"rationale": "First, the post has a conversational, forum-style voice with personal markers ('CMV', rhetorical questions, asides) typical of human online posts. Second, it shows idiosyncratic errors and informal spellings ('incidentialally'), inconsistent punctuation, and emotional emphasis that suggest spontaneous... | {
"rationale": "First, the post has a conversational Reddit style (CMV, rhetorical questions, personal voice) and includes community-specific markers that suggest human posting. Second, there are multiple typographical errors and inconsistent word choices ('Incidentialally', 'comfrt') plus uneven sentence-level issue... | {
"rationale": "First, the post has a conversational Reddit style (CMV, rhetorical questions, colloquial phrasing) and shows personal voice and emotional engagement typical of human forum posts. Second, there are multiple typos and inconsistent word choices ('Incidentialally', 'comfrt') suggesting spontaneous human t... | {
"rationale": "First, the post has a highly conversational, forum-style voice with personal markers ('CMV', rhetorical questions, emotional emphasis) typical of human online posts. Second, there are multiple typos and inconsistent spellings ('incidentialally', 'comfrt') suggesting spontaneous human typing rather tha... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 651 |
human-2180 | human | I believe that Transgenderism is, in fact, a mental disorder, akin to schizophrenia or depression. In fact, it was classified as such by the American Psychiatric Association until less than a year ago; that is part of why I hold this view. Furthermore, there is no physical manifestation of Transgenderism; it is all in ... | NA | cmv | {
"rationale": "First, the passage uses a personal, conversational voice with specific temporal markers ('less than a year ago') and anecdotal framing that is typical of human online posts. Second, it shows small stylistic inconsistencies and colloquial choices (e.g., 'per se', contractions) rather than uniformly pol... | {
"rationale": "First, the passage uses a conversational, personal voice with contractions and rhetorical markers (e.g., 'I believe') that match human forum posts. Second, it shows small stylistic inconsistencies and colloquialisms ('In fact,' 'per se') suggesting spontaneous human writing rather than polished AI out... | {
"rationale": "First, the passage has a conversational, personal voice with colloquial transitions (e.g., 'In fact,' 'Furthermore') typical of human forum posts. Second, it shows small factual imprecision and subjective hedging ('until less than a year ago'; 'I believe') suggesting spontaneous human opinion rather t... | {
"rationale": "First, the passage has a conversational, personal voice with colloquial transitions (e.g., 'In fact,' 'Furthermore') typical of human online posts. Second, it shows small factual imprecision and emotional bias (sudden shift from claim to justification) suggesting spontaneous human writing rather than ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 316 |
human-951 | human | In pretty much any other situation, it's nearly universally agreed that an action that would otherwise be immoral does not suddenly become moral because the person you did it to did something bad first. I see no reason why this principle should not be applied to the case in which you attack someone who attacked you. So... | NA | cmv | {
"rationale": "First, the passage uses polished, balanced language and generic transitional phrases ('In pretty much any other situation...'), a common trait of AI-generated text aiming for neutrality. Second, it avoids personal idiosyncrasies or concrete details and relies on abstract hedging ('some might argue', '... | {
"rationale": "First, the passage uses polished, balanced language and smooth transitions with few personal idiosyncrasies, a pattern common in AI-generated text. Second, it avoids concrete examples and relies on abstract hedging ('some might argue') while asserting strong normative claims (e.g., 'the right to life.... | {
"rationale": "First, the passage uses polished, balanced language and smooth transitions with few personal idiosyncrasies, a pattern common in AI-generated text. Second, it avoids concrete examples and relies on abstract hedging ('some might argue') while making bold absolutist claims (e.g., 'the right to life... t... | {
"rationale": "First, the passage uses polished, balanced language and smooth transitions without personal idiosyncrasies, suggesting algorithmic consistency rather than a distinct human voice. Second, it avoids concrete examples and relies on abstract hedging ('some might argue'), a common trait of AI-generated tex... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 324 |
human-1338 | human | Whenever a person of the male gender argues on the side of feminism or other forms of equality, they are labeled as being a "white knight" or a "social justice warrior." They are also accused of having self-profiting motivations rather than genuine motivations. For example, they may be accused of "stirring the pot" or ... | NA | cmv | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, anecdotal tone typical of human online posts. Second, it shows idiosyncratic phrasings, rhetorical questions, and mild inconsistencies that suggest spontaneous human composition rather t... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, anecdotal tone typical of human online posts. Second, it shows idiosyncratic phrasings, rhetorical questions, and small inconsistencies (e.g., 'How do we know that?') suggesting spontane... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed link placeholders) and colloquial, context-rich references ('Bob Chipman... or David Willis') consistent with copied human forum text. Second, there are small stylistic inconsistencies, rhetorical questions, and mil... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, named posters) and colloquial, context-dependent phrasing typical of human online discussion. Second, there are small inconsistencies and stylistic idiosyncrasies (repetition, casual tone, bracketed link placeholders) suggest... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 468 |
human-963 | human | Yesterday, it was announced that [the US Marine Corps. was delaying it's female fitness plan] after less than half its female recruits could do the minimum of three pull-ups. Then there's the story of [Wendy Tapia] , who despite failing a required FDNY running test five times (a 1.5 mile run in 12 minutes without gear)... | NA | cmv | {
"rationale": "First, the post shows forum-specific markers ('CMV', 'Edit:', 'delta') and a conversational, self-correcting tone typical of human online discussion. Second, it contains idiosyncratic errors and stylistic quirks (misused ellipses, inconsistent spacing, bracketed asides) that AI rarely produce so natur... | {
"rationale": "First, the post shows forum-specific markers ('CMV', 'Edit:', 'delta') and a conversational, self-correcting tone typical of human online discussion. Second, it contains idiosyncratic errors and stylistic quirks (misused ellipses, inconsistent spacing, bracketed link placeholders) suggesting casual hu... | {
"rationale": "First, the post shows forum-specific markers ('CMV', 'Edit:', 'delta') and a conversational, self-correcting tone typical of human online discussion. Second, it contains idiosyncratic errors and stylistic quirks (misused ellipses, inconsistent spacing, bracketed link placeholders) suggesting casual hu... | {
"rationale": "First, the post displays a highly conversational, forum-specific voice with markers like 'CMV', 'Edit:', bracketed links, and personal deltas which are typical of human Reddit posts. Second, it contains idiosyncratic phrasings, emotional hedging ('I would never advocate...'), and uneven structure sugg... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 571 |
human-858 | human | I suspect that rMensRights will be the more controversial of these choices. Anyways, I should clarify what I mean. These kinds of subs "leak" into the rest of reddit, while tainting it's image. Most of the people I know have hear of reddit either as as home for pedophiles, or a sexist racist mess. Ignoring whether thei... | NA | cmv | {
"rationale": "First, the post shows colloquial, context-specific markers like 'Anyways' and 'CMV' and uses profanity and personal asides in a way typical of human online comments. Second, there are small typos and repetition ('as as') and inconsistent grammar that suggest spontaneous human typing rather than polish... | {
"rationale": "First, the post uses casual, forum-specific language ('Anyways', 'CMV') and shows typographical slips ('as as') consistent with spontaneous human typing. Second, it displays strong personal voice, profanity, and inconsistent grammar/usage ('it's image' followed by misuse of 'its') suggesting human err... | {
"rationale": "First, the post uses casual, forum-specific language and slang ('Anyways', 'CMV') and shows typographical slips ('as as') consistent with spontaneous human typing. Second, there are inconsistent grammar choices and colloquial repetition ('it's image' then 'I should clarify...') suggesting human incons... | {
"rationale": "First, the post shows colloquial, context-specific Reddit behavior (use of 'as as' and 'CMV') and small typos/grammatical slips ('have hear of reddit') typical of spontaneous human typing. Second, it contains strong personal voice, profanity, and emotionally nuanced uncertainty ('I suspect...'), sugge... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 273 |
human-173 | human | Subjective sports cheapen Olympic medals, in an objective sport, the first person across the line of the team with the most goals wins. In a subjective sport, judges cannot help but be biased towards people that they perceive as being good, or from countries that they like. It's a psychological thing. On top of that, E... | NA | cmv | {
"rationale": "First, the passage shows a casual, conversational tone with personal asides ('It's a psychological thing') and inconsistent formatting/typos (e.g., 'GymnasticsRythmic', missing space after '-'), which are typical of quick human writing. Second, it contains specific cultural references and idiosyncrati... | {
"rationale": "First, the passage shows colloquial phrasing, inconsistent punctuation/capitalization (e.g., 'List of subjective Olympic sports...' with no list) and typographical errors ('GymnasticsRythmic') typical of quick human typing. Second, it contains personal voice and opinionated asides ('It's a psychologic... | {
"rationale": "First, the passage shows casual, conversational phrasing and idiosyncratic errors (e.g., 'GymnasticsRythmic') more typical of quick human typing than polished AI output. Second, it contains personal voice ('It's a psychological thing'), inconsistent structure, and list formatting suggesting human note... | {
"rationale": "First, the passage shows abrupt topic shifts, mechanical transitions ('On top of that'), and inconsistent tone (serious then slang), patterns common in AI-generated text. Second, there are token-level oddities—stray spaces before punctuation, concatenated words ('GymnasticsRythmic'), and repetitive 'i... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 275 |
human-1770 | human | EDIT: Dear guitarists of Reddit, thank you for some amazing comments! Although I have realised my view is far too vague to be debated with good results, if I were to re-submit the same sort of view to this sub-reddit I would make it a lot more precise. Once again, a rude comment won't get you anything but reported. The... | NA | cmv | {
"rationale": "First, the passage contains forum-specific markers (EDIT tag, subreddit conventions, moderator footnote with bracketed links) and casual, idiosyncratic phrasing typical of human posts. Second, there are small inconsistencies, colloquialisms ('tbh'), and uneven punctuation/spacing that suggest spontane... | {
"rationale": "First, the passage contains highly contextual, community-specific markers (EDIT tag, subreddit conventions, moderator footnote with bracketed links) that match real Reddit posts rather than generic AI output. Second, it shows idiosyncratic phrasings, small grammatical slips, and conversational tone ('... | {
"rationale": "First, the passage contains highly contextual, community-specific markers (EDIT tag, subreddit conventions, moderator footnote with bracketed links) consistent with a copied Reddit post rather than AI-generated generic text. Second, there are small stylistic inconsistencies, colloquialisms ('tbh'), an... | {
"rationale": "First, the passage contains highly contextual, community-specific markers (EDIT tag, subreddit conventions, moderator footnote) consistent with a copied Reddit post rather than AI-generated generic text. Second, it shows human-like idiosyncrasies — casual shorthand (tbh), uneven punctuation, personal ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 515 |
human-1420 | human | Reposting is probably one of, if not the most often bemoaned actions on Reddit. Users are incredibly quick to point out if content has been posted before, especially in TIL, either as some strange form of boast or by negating the OP. Frequently I look into the comments section for more info or commentary on the post, o... | NA | cmv | {
"rationale": "First, the passage uses colloquial, context-specific markers (TIL, OP, CMv) and a personal voice typical of forum posts. Second, it shows nuanced opinion, rhetorical questions, and mild stylistic inconsistencies that humans commonly produce whereas AI outputs often smooth or overexplain. Moreover, the... | {
"rationale": "First, the passage uses colloquial, context-specific forum language (e.g., 'Cmv', 'OP', 'TIL') and shows personal, opinionated nuance typical of human posts. Second, it contains small stylistic inconsistencies and rhetorical questions ('Or is there a benefit?') suggesting spontaneous human writing rat... | {
"rationale": "First, the passage uses colloquial, context-specific forum language (e.g., 'Cmv', 'OP', 'TIL') and shows personal, idiosyncratic phrasing typical of human posts. Second, it contains minor stylistic inconsistencies and rhetorical questions ('Or is there a benefit?') suggesting spontaneous human thought... | {
"rationale": "First, the passage contains forum-specific conventions (TIL, OP, CMv) and a conversational, idiosyncratic tone typical of human casual posts. Second, it shows small stylistic inconsistencies and personal hedging ('I would be interested...') rather than polished, overly generic AI output. Moreover, the... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 326 |
human-248 | human | (quick note: I am in general OK with people using seximracismwhat-have-you jokingly. In specific instances in can be bad, but that isn't really what this CMV is about) I often see people who are otherwise in favor of gender equality use the phrase 'man up' to tell people that they are being a coward and need to stop be... | NA | cmv | {
"rationale": "First, the passage has a conversational, forum-specific voice with typos and idiosyncratic phrasings ('in general OK', 'seximracismwhat-have-you') that match human casual posting. Second, it shows personal context, emotional nuance, and uneven sentence-level errors rather than polished, consistently s... | {
"rationale": "First, the post uses casual, forum-specific phrasing ('CMV') and includes personal context ('I am in general OK...'), which matches human conversational posting. Second, there are small typos and idiosyncrasies ('in specific instances in can be bad', 'seximracismwhat-have-you') suggesting spontaneous ... | {
"rationale": "First, the post uses casual, forum-specific language ('CMV') and personal context ('I am in general OK...'), which matches human online posting patterns. Second, there are typographical slips and uneven punctuation ('in specific instances in can be bad') suggesting spontaneous typing rather than polis... | {
"rationale": "First, the post uses casual, forum-specific language ('CMV') and personal context ('I am in general OK...'), suggesting a human conversational tone. Second, there are typographical slips and uneven punctuation ('in specific instances in can be bad') typical of quick human typing rather than polished A... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 274 |
human-2091 | human | Pretty self-explanatory title. I don't think Nightmares 3 - 6 (Freddy's Dead) were meant to scare people, they were slasher comedies. In Dream Warriors (3) the kills were comedic in nature (including my all-time favorite: "Welcome to Prime Time, Bitch!, and apart from Kristen, none of the kids were especially sympathet... | NA | cmv | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, idiosyncratic voice consistent with human online posting. Second, it shows personal taste, casual digressions, and inconsistent punctuation/parentheses that are typical of spontaneous hu... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, anecdotal tone typical of human online posts. Second, there are small inconsistencies, colloquial phrasings, and copy-paste artifacts (mismatched quotes, odd spacing) suggesting manual e... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, anecdotal tone typical of human online posts. Second, there are small inconsistencies, colloquial phrasings, and copy-paste artifacts (mismatched quotes, spacing, and explicit 'Not Inclu... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, anecdotal tone consistent with human online posting. Second, there are idiosyncratic errors and informal phrasings (e.g., mismatched quotes, missing spaces) suggesting spontaneous human ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 518 |
human-1422 | human | I know there's a lot that goes along with this (that I will try to address sufficiently below)... but for a tl;dr my reasoning is that if other nations don't hate us, we won't have to spend ridiculous amounts of money on the military because no big conflict will arise. Now I know this is an extremely complicated matter... | NA | cmv | {
"rationale": "First, the passage shows a conversational, idiosyncratic voice with colloquialisms ('tl;dr'), contractions, and typographical slips ('mayve', 'millitary') that are typical of human typing. Second, it contains personal hedging, moral ambivalence, and uneven organization—narrative digressions and self-c... | {
"rationale": "First, the post uses a highly conversational, forum-style voice with personal hedging ('I know', 'I've always thought') and includes casual typos/colloquialisms ('mayve', 'tl;dr') typical of human online writing. Second, it shows nuanced uncertainty, self-correction, and specific cultural references (... | {
"rationale": "First, the post shows a conversational, idiosyncratic voice with colloquialisms ('tl;dr'), contractions ('mayve'), and typographical slips ('millitary') typical of human forum writing. Second, it contains personal hedging, uneven structure, and specific cultural references (Brazil, 400 billion) sugges... | {
"rationale": "First, the post shows a highly conversational, idiosyncratic voice with personal hedging ('I know', 'I've always thought') and forum-specific cues (tl;dr, parentheses, slang) typical of human online writing. Second, it contains small errors and stylistic inconsistencies (misspelling 'mayve', casual co... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 572 |
human-2184 | human | I'm not confident that we can avert global warming. If we all acted in unison, today, we could almost certainly fix the problem. I don't know we will. I think even if we do fix this problem, we've only delayed the greater problem (habitability problems) by a few millennia. As long as we're on Earth, we're vulnerable. T... | NA | cmv | {
"rationale": "First, the post uses a casual, forum-specific voice ('Well, go on, CMV') and includes community conventions like 'etc, etc' and personal hedging ('I'm not confident...'), which are typical of human online posts. Second, it shows idiosyncratic phrasings and small grammatical slips ('economically places... | {
"rationale": "First, the post uses a casual, conversational tone with forum-specific shorthand ('CMV') and personal hedging ('I'm not confident...'), matching typical human online posting. Second, it shows idiosyncratic phrasings ('economically places are terrible', 'etc, etc') and small grammatical slips that huma... | {
"rationale": "First, the post uses a casual, conversational tone with forum-specific shorthand ('CMV') and personal hedging ('I'm not confident...'), patterns common in human online posts. Second, it shows idiosyncratic phrasings and small grammatical slips ('economically places are terrible', 'etc, etc') suggestin... | {
"rationale": "First, the post uses a highly conversational, forum-specific voice ('Go on, CMV') and personal uncertainty markers ('I'm not confident...'), suggesting a human Reddit-style poster. Second, it shows colloquial fillers ('etc, etc'), uneven phrasing ('economically places are terrible'), and small stylist... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 343 |
human-1471 | human | I don't understand why people support the Palestinian cause. Palestinian nationalism is baseless, and Palestinian groups with nationalistic aspirations use extremely violent means to persuade the Israeli state to give Palestinian people more rights. How can it be said that they deserve it after all of the murders they ... | NA | cmv | {
"rationale": "First, the post uses a personal, conversational voice with direct questions and emotional engagement ('Can someone please explain...'), which is typical of human online posts. Second, it shows nuanced, context-aware beliefs (support for Palestinians' rights while opposing violence) and avoids generic,... | {
"rationale": "First, the post uses a personal, conversational tone with direct questions and emotional engagement ('Can someone please explain...'), which matches human online posting patterns. Second, it shows small stylistic inconsistencies (contraction 'don't', emphatic 'IS' in parentheses, comma placement) sugg... | {
"rationale": "First, the post uses a personal, conversational tone with direct questions and emotional engagement ('Can someone please explain...'), which is typical of human online posts. Second, there are small stylistic inconsistencies (all-caps 'IS', casual phrasing) suggesting spontaneous typing rather than po... | {
"rationale": "First, the post uses a highly conversational, forum-style voice with personal uncertainty and direct appeal for persuasion (\"Can someone please explain...\"), which matches human online behavior. Second, it shows emotional investment, colloquial emphasis (ALL CAPS \"IS\") and small stylistic inconsis... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 241 |
human-390 | human | As the title says, I've recently fallen in with a group of people who love to drink and party together. By that I mean that more or less the same group of people get together and have drinks at one of their houses. Every weekend they're all so excited to get together for drinks and all I can think of is all the other m... | NA | cmv | {
"rationale": "First, the passage contains highly contextual, community-specific markers (CMV, moderator footnote, bracketed links, edit note) and colloquial phrasing typical of human forum posts. Second, there are small inconsistencies, typos ('spending time them'), and conversational redundancies that suggest spon... | {
"rationale": "First, the passage contains highly contextual, community-specific markers (CMV, moderator footnote, bracketed links, edit note) and conversational tone typical of human forum posts. Second, there are small typos, inconsistent spacing, and colloquial phrasings ('I like spending time them') suggesting s... | {
"rationale": "First, the passage contains highly contextual, community-specific markers (CMV, moderator footnote, bracketed links, edit note) and conversational tone typical of human forum posts. Second, there are small typos, colloquial phrasings ('throwaway', 'Productive really wasn't the word') and uneven gramma... | {
"rationale": "First, the passage contains highly contextual, community-specific markers (CMV, moderator footnote, bracketed links, edit note) consistent with a copied Reddit post, which is typical of human forum content. Second, it shows casual typos, inconsistent spacing, emotive emphasis (e.g., MUCH), and convers... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 466 |
human-1756 | human | 1. The presenter overdoes the "you have to stop this right now" way of using the buzzer even when it's not necessary. 2. She laughs too much at the jokes it almost feels fake. 3. 3 ads in a 30-minute show compared to 1 in the original British show, one of which is just before the credits. 4. Having guest stars and leav... | NA | cmv | {
"rationale": "First, the text uses casual, idiosyncratic phrasing ('That's all I can think of for now') and small inconsistencies (double space, uneven bullet depth) typical of quick human writing. Second, it shows personal subjective judgments and specific cultural reference (British show) with conversational tone... | {
"rationale": "First, the text uses casual, conversational phrasing and personal markers (e.g., 'That's all I can think of for now') typical of human forum posts. Second, it shows small inconsistencies and idiosyncratic choices ('3 ads in a 30-minute show' with a missing hyphen) suggesting spontaneous typing rather ... | {
"rationale": "First, the text uses casual, conversational phrasing and personal markers ('That's all I can think of for now') typical of human forum posts. Second, there are small inconsistencies and colloquial slips (repeated '3', odd comma placement) suggesting spontaneous typing rather than polished AI output. M... | {
"rationale": "First, the text uses casual, first-person phrasing ('That's all I can think of for now') and includes concrete personal observations (laughs, forgetting to assign points) suggesting lived viewing rather than generic AI output. Second, there are small stylistic inconsistencies and colloquialisms ('even... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 328 |
human-57 | human | I have not had an easy life. My parents were abusive, and as a result I developed PTSD. Recently the psych has started suggesting bipolar, and it fits. I've struggled with eating disorders, anxiety, and self harm. I've lost everyone I cared about, including my grandparents (who were more parents to me than my regular p... | NA | cmv | {
"rationale": "First, the post shows a highly personal, emotionally nuanced voice with specific lived details (PTSD, eating disorders, grandparents) and community-specific shorthand ('CMV') that match genuine Reddit submissions. Second, there are typographic idiosyncrasies and uneven spacing ('IQjust the facts', '50... | {
"rationale": "First, the post shows highly specific personal details, emotional nuance, and forum-specific markers ('CMV', 'unattractive') that match genuine human online behavior. Second, there are small typographic inconsistencies and colloquial phrasings ('I've lost everyone... including my grandparents...') whi... | {
"rationale": "First, the post displays highly specific personal details, emotional nuance, and forum-specific markers ('CMV', 'unattractive') consistent with a real Reddit user's submission. Second, there are small typographic idiosyncrasies and uneven phrasing ('IQjust the facts', '50 peoplehas') typical of human ... | {
"rationale": "First, the post displays highly specific personal details, emotional rawness, and forum-specific markers ('CMV', 'unattractive') consistent with a genuine Reddit submission. Second, there are typographic idiosyncrasies and uneven spacing ('IQjust the facts'), casual phrasing, and moral ambiguity that ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 410 |
human-1831 | human | First off I'll admit that I'm biased by Western liberal media - especially the likes of Jon Stewart, Stephen Colbert, Bill Maher, Charlie Brooker and numerous documentaries from the liberal side of the spectrum, particularly Adam Curtis, in his "Power of Nightmares"; and numerous documentaries on inequality and the glo... | NA | cmv | {
"rationale": "First, the post shows a conversational, self-correcting forum-style voice with explicit 'Edit:' and personal asides that are typical of human online posts. Second, it contains idiosyncratic choices—typos ('i' for lowercase i'), inconsistent capitalization, repetition, and emotionally nuanced bias—that... | {
"rationale": "First, the post shows a conversational, self-referential tone with personal admissions ('First off...'), an 'Edit:' note, and a signed recommendation to read a specific document, patterns common in human forum posts. Second, there are small inconsistencies, colloquial emphases (e.g., 'i DO NOT...') an... | {
"rationale": "First, the post shows a conversational, forum-style voice with personal admissions ('First off...'), an 'Edit:' note, and a direct invitation to respond, patterns common in human online posts. Second, there are small inconsistencies, colloquial emphases (e.g., 'i DO NOT...') and a pasted link prompt s... | {
"rationale": "First, the post shows a highly personal, conversational Reddit-style voice with explicit 'Edit:' and first-person admissions that match human forum behavior. Second, it contains idiosyncratic emphases (all-caps, repeated words, casual typos like 'i' for an em dash) suggesting spontaneous human typing ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 788 |
human-436 | human | When I debate whether or not humans have free will, someone often brings up blame. "If there is no free will, we can't hold anyone accountable for their actions." I believe that the justice system should not be designed to punish or to exact revenge. We can still impose penalties on criminals for two separate purposes.... | NA | cmv | {
"rationale": "First, the post uses a conversational, forum-style voice with personal markers ('When I debate...') and direct appeals ('Change my view'), which matches human online posting patterns. Second, it shows idiosyncratic phrasings, small stylistic inconsistencies, and emotional hedging ('not that I wouldn't... | {
"rationale": "First, the post uses a conversational, forum-style voice with personal markers ('When I debate...') and direct appeals ('Change my view'), which matches human online posting patterns. Second, it shows small stylistic inconsistencies, colloquial phrasings ('the horse has been slaughtered') and minor gr... | {
"rationale": "First, the post uses a conversational, forum-style voice with personal appeals ('Change my view') and colloquial contractions, which is typical of human online posts. Second, it shows small stylistic inconsistencies and idiosyncratic phrasings ('a persons behavior', 'not that horse has been slaughtere... | {
"rationale": "First, the post uses a conversational, forum-style voice with personal appeals ('Change my view') and colloquial phrasings ('that horse has been slaughtered'), suggesting spontaneous human posting. Second, it shows small stylistic inconsistencies and emotional engagement rather than polished, evenly b... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 330 |
human-1254 | human | My fiancee tells me that she has need to make people happy. The problem is that it's detrimental to her health, and I keep trying to tell her that. What I don't understand is that how is it possible for people to feel this innate urge to help people when it causes them to feel worse about themselves? I first started wo... | NA | cmv | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a personal anecdote with idiosyncratic phrasing ('My fiancee') typical of human online posts. Second, it shows uneven grammar, repetition, and emotional nuance ('I keep trying to tell her that') suggestin... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a personal anecdote with emotional nuance that match human online posting. Second, it shows idiosyncratic phrasing, repetition ('selfishly selfless'), and a minor grammatical slip ('that she has need' ins... | {
"rationale": "First, the passage shows forum-specific markers (CMV, moderator footnote, bracketed links) and a personal, anecdotal voice consistent with human online posting. Second, it contains idiosyncratic phrasings, minor grammatical slips ('that she has need to make people happy'), and conversational repetitio... | {
"rationale": "First, the passage contains highly contextual, forum-specific markers (CMV, moderator footnote, personal anecdote) and casual phrasing consistent with human online posts. Second, it shows idiosyncratic errors and stylistic inconsistencies (typos, tense shifts, repetition) typical of spontaneous human ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 515 |
human-943 | human | I'm in no way against the ideas of wanting to explore the world, find out how things work, discover more about ourselves and better our own lives through technology. Nevertheless, I feel that placing so much trust in the idea that all of our questions can be answered and 'proven' (compared to being content with leaving... | NA | cmv | {
"rationale": "First, the passage contains forum-specific markers ('tl;dr', 'ucaw81') and casual, idiosyncratic phrasings typical of human online posts. Second, there are typographic errors and inconsistent spacing ('mentalpsychological'), plus personal asides and emotive language ('I may not reply... Good talk gang... | {
"rationale": "First, the passage contains forum-specific markers ('tl;dr', 'ucaw81') and casual conversational cues that match human online posting. Second, there are typographic slips and inconsistent spacing ('mentalpsychological'), plus personal asides and emotive language suggesting spontaneous human typing rat... | {
"rationale": "First, the passage shows highly conversational, forum-style traits — explicit mentions of Reddit handles (ucaw81), 'Good talk gang!', and repeated 'tl;dr' markers consistent with human online posts. Second, there are typographic idiosyncrasies and casual errors (e.g., 'mentalpsychological' run togethe... | {
"rationale": "First, the passage contains highly idiosyncratic forum-style markers (tl;dr, ucalw81, thanks, read replies) and casual typos/spacing issues that are typical of human posts. Second, there is a mix of personal voice, emotional nuance, and uneven editing (bracketed note, missing space after a period) sug... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 399 |
human-2303 | human | Let me make my view clear here. I believe that things like feminism and men's rights comeswill come in waves. First wave feminism was basically "We really have no rights, let's change that." Second wave feminism is "Hey, we have a lot more than we had, but there are still some pretty big, obvious inequalities that need... | NA | cmv | {
"rationale": "First, the passage shows a highly conversational, idiosyncratic voice with typos, colloquialisms, and self-deprecating asides (e.g., 'Sorry if the writing is weird') typical of human forum posts. Second, it contains personal anecdotes, inconsistent punctuation/capitalization ('comeswill', 'Second wave... | {
"rationale": "First, the passage shows a highly conversational, forum-style voice with personal anecdotes, hedging, and explicit self-reflection that matches human online posts. Second, it contains typographical errors and colloquialisms ('comeswill', 'poop') plus inconsistent grammar/structure suggesting spontaneo... | {
"rationale": "First, the passage shows a highly conversational, forum-style voice with personal anecdotes, hedging, and explicit apology ('Sorry if the writing is weird'), traits common in human online posts. Second, it contains typographical errors and uneven grammar ('comeswill', 'in text') suggesting casual huma... | {
"rationale": "First, the passage shows a highly conversational, forum-style voice with personal anecdotes, hedging, and explicit apology ('Sorry if the writing is weird'), traits common in human posts. Second, it contains typographical errors and uneven grammar ('comeswill', 'in text') suggesting casual human typin... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 531 |
human-889 | human | Some prefaces (sad that I have to do this lest the pitchfork mob burn me at the stake): Animal abuse does get me riled up. The whole fiasco of yesterday with the abused poodle puppy on the dancing puppy gif (turned out to be a different dog) is what got me thinking about this. I watched the whole 15 minute video of the... | NA | cmv | {
"rationale": "First, the passage shows a highly idiosyncratic, emotionally charged voice with colloquialisms, profanity, digressions, and self-corrections that read like spontaneous human writing. Second, it contains small inconsistencies, bracketed placeholders ('[isught]'), inconsistent capitalization, and rhetor... | {
"rationale": "First, the passage shows a highly idiosyncratic, emotionally raw voice with colloquialisms, profanity, and inconsistent punctuation/parentheses that are typical of spontaneous human writing. Second, it contains specific personal details, rhetorical questions, and tentative philosophical markers (e.g.,... | {
"rationale": "First, the passage shows a highly idiosyncratic, emotionally raw voice with colloquialisms, profanity, digressions, and self-corrections that read like spontaneous human writing rather than polished AI output. Second, there are small inconsistencies, bracketed placeholder tokens ('[isught]'), uneven p... | {
"rationale": "First, the passage shows a highly idiosyncratic, emotionally raw, and conversational voice with slang, profanity, and self-deprecating asides that read like spontaneous human writing. Second, it contains uneven grammar, personal tangents, bracketed placeholder 'isought', and rhetorical questions sugge... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 938 |
human-1410 | human | Though obviously there was a famine with the potatoes that was only part of the problem. The British oppression of the Irish people though the corn laws (not allowing wheat and other staple food items from other countries to help the Irish) and poorly planned relief efforts are all the British's fault. They basically r... | NA | cmv | {
"rationale": "First, the post uses forum-specific shorthand ('CMV') and a personal, emotionally charged tone common in human online discussions. Second, it shows colloquial phrasings and minor inconsistencies (e.g., 'British's' and repeated 'though') suggesting spontaneous human writing rather than polished AI outp... | {
"rationale": "First, the post uses forum-specific shorthand ('CMV') and a personal, emotionally charged tone common in human online posts. Second, it shows colloquial phrasings ('all the British's fault'), inconsistent punctuation/spacing, and repetition that suggest spontaneous human typing rather than polished AI... | {
"rationale": "First, the post uses forum-specific shorthand ('CMV') and a personal, emotionally charged tone common in human online posts. Second, it shows colloquial phrasings ('all the British's fault'), inconsistent punctuation/spacing, and repetition that suggest spontaneous human typing rather than polished AI... | {
"rationale": "First, the post uses forum-specific shorthand ('CMV') and a personal, emotionally charged tone common in human online posts. Second, it shows colloquial phrasings ('all the British's fault'), inconsistent punctuation/spacing, and repetition suggesting spontaneous human typing rather than polished AI o... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 227 |
human-765 | human | Let me start off by saying that I think sampling and photography are neither cheating or unartistic. In fact, some of my favorite pieces of art either contain samples or are photos. Now, I often hear people say that sampling is cheating because the artist didn't make the sample but rather "stole" it and put it into the... | NA | cmv | {
"rationale": "First, the post shows forum-specific markers (CMV header, moderator footnote, bracketed links) and a conversational, context-aware tone typical of human online posts. Second, it contains idiosyncratic errors and informal choices (e.g., 'andor', casual phrasings, rhetorical questions) that AI rarely pr... | {
"rationale": "First, the post shows forum-specific markers (CMV header, moderator footnote, bracketed link placeholders) and a conversational, context-aware tone typical of human online posts. Second, it contains idiosyncratic errors and informal choices (e.g., 'andor', casual phrasings, uneven punctuation) that ar... | {
"rationale": "First, the post shows forum-specific markers (CMV header, moderator footnote, bracketed link placeholders) and a conversational, context-aware tone typical of human online posts. Second, it contains idiosyncratic errors and informal choices (e.g., 'andor', casual phrasings, uneven punctuation) suggest... | {
"rationale": "First, the post shows forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, context-aware tone typical of human online posts. Second, it contains idiosyncratic errors and informal choices (e.g., 'andor', casual phrasings, uneven punctuation) suggesting spontaneous hum... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 901 |
human-1228 | human | To be clear, I don't think polygamy would be the right course of action for me. But I do believe that if both men and women have the right to marry as many or as few people as they want in a society that does not in any way provide economic incentives for marriage, then there is no problem with that, in my eyes. Can an... | NA | cmv | {
"rationale": "First, the passage uses conversational, context-specific phrasing ('To be clear', 'Can anybody change my view') and shows personal nuance and uncertainty typical of human online dialogue. Second, it contains small stylistic idiosyncrasies (e.g., 'in a society that does not in any way provide economic ... | {
"rationale": "First, the post uses conversational, context-specific phrasing ('To be clear...') and a direct interpersonal query ('Can anybody change my view?'), which matches human online discussion. Second, it shows personal hedging (I don't think...), moral nuance, and a balanced, tentative stance rather than ov... | {
"rationale": "First, the passage uses conversational, context-specific phrasing ('To be clear...') and a direct interpersonal query ('Can anybody change my view?'), which is typical of human online posts. Second, it shows tentative, personal stance and nuance (don't think polygamy is right for me but believe under ... | {
"rationale": "First, the passage uses balanced, generic phrasing ('To be clear...') and hedging ('I don't think...'), patterns common in AI-generated conversational replies. Second, it lacks personal detail or concrete examples and relies on abstract hypotheticals (both men and women having the right to marry), sug... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 228 |
human-2126 | human | The worst of words are used to offend because they work. If people looked at "Faggot" and "Nigger" as the unimaginative word choices that they are people would laugh at the people saying it as much as "poop head" I completely understand being offended when someone is spewing hate. I don't understand why people allow ce... | NA | cmv | {
"rationale": "First, the passage shows a raw, colloquial voice with typos, inconsistent punctuation, and idiosyncratic phrasings (e.g., 'Sounds we make with our mouth') typical of spontaneous human writing. Second, it contains personal anecdotes, emotional nuance, and uneven argumentation ('My best friend is black.... | {
"rationale": "First, the passage shows a raw, emotionally charged voice with colloquial phrasings, spelling/grammar slips (e.g., 'speach', 'perpose') and inconsistent punctuation that are typical of spontaneous human writing. Second, it contains personal anecdotes, contradictory self-reflection ('I completely under... | {
"rationale": "First, the passage shows a raw, emotionally charged voice with colloquial slang, inconsistent punctuation, and typographical errors (e.g., 'speach') typical of spontaneous human writing. Second, it contains personal anecdotes, moral inconsistency, and rhetorical questions suggesting lived experience r... | {
"rationale": "First, the passage shows a raw, emotionally charged forum-style voice with typos, inconsistent punctuation, and colloquial phrasing (e.g., 'Sounds we make with our mouth') suggesting spontaneous human typing. Second, it contains personal anecdotes, moral inconsistency, and rhetorical questions that AI... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 351 |
human-533 | human | I think the end game is to homogenize unique cultures and even out markets so big corporation will have more control. It feels like they treat national cultures as commodities, especially indigenous ones. | NA | cmv | {
"rationale": "First, the message is brief, conversational, and uses first-person phrasing ('I think') and contractions, matching casual human forum posts. Second, it shows minor stylistic idiosyncrasies (e.g., 'so big corporation' without 'the') and a subjective, experience-driven tone rather than polished, neutral... | {
"rationale": "First, the short, conversational tone and use of first-person phrasing ('I think') suggest a casual human voice rather than typical AI's often more formal or hedged output. Second, there are small stylistic idiosyncrasies — lack of excessive polishing, contractions, and colloquial word choices like 'h... | {
"rationale": "First, the message is brief, conversational, and uses first-person phrasing ('I think') which is common in human online replies. Second, there are small stylistic idiosyncrasies—'big corporation' lacking a hyphen and casual sentence breaks—that suggest spontaneous human typing rather than polished AI ... | {
"rationale": "First, the text uses casual, first-person phrasing ('I think') and a conversational tone typical of human forum posts. Second, it shows minor stylistic slips (missing apostrophe in 'big corporation' and colloquial word choice) suggesting spontaneous human writing rather than polished AI output. Moreov... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 188 |
human-349 | human | Hi CMV, let me present myself, I'm a 17 year old pianist, have been taking lessons since 10, and only till now I have wanted to have a serious music career as a composer more than a interpreter. But I feel that the fact that I have almost no money to spend on my musical education is making me think I may not be able to... | NA | cmv | {
"rationale": "First, the post shows a highly personal, colloquial forum voice with specific lived details (age, family finances, local conservatory, Chopin's Etude op10 no3) typical of human user posts. Second, it contains typographical errors, inconsistent punctuation, tense shifts, and informal phrasings ('change... | {
"rationale": "First, the post shows a highly personal, conversational Reddit-style voice with specific cultural details (pianist, 17 years old, Mexico) and idiosyncratic phrasings that are typical of human writing. Second, it contains typographical errors and inconsistent word choices ('feel' for 'felt', 'composeri... | {
"rationale": "First, the post shows a highly personal, conversational Reddit-style voice with specific cultural details (Mexico, local conservatory, Chopin's Etude op10 no3) that AI rarely replicate so naturally. Second, it contains typographical errors, tense shifts, run-on sentences, and informal phrasings ('any ... | {
"rationale": "First, the post shows a highly personal, colloquial forum-style voice with specific lived details (age, location, family finances) typical of human user posts. Second, it contains multiple typographical errors, inconsistent punctuation/capitalization, and idiosyncratic phrasings ('an scholarship', mis... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 567 |
human-267 | human | Syria: another war for oil, being perpetrated to enrich American military and infrastructure contractors, and to serve as a proxy war to weaken Russia and Iran (Syria's biggest allies) in order to enrich international banking interests (who are threatened by Russia's and the East's expanding economic power) and empower... | NA | cmv | {
"rationale": "First, the passage shows a distinct personal voice, rhetorical questions, colloquialisms ('butt-load') and emotional engagement typical of human forum posts. Second, it contains typographical errors and inconsistent formatting/parentheses (e.g., 'WahhabistsSalafists'), suggesting casual human typing r... | {
"rationale": "First, the piece shows a distinct personal voice, rhetorical questions, colloquialisms ('butt-load'), and emotional appeals ('spilt') that are typical of human forum posts. Second, there are typographic and factual inconsistencies (e.g., 'WahhabistsSalafists' and missing '%' after '78') suggesting cas... | {
"rationale": "First, the passage shows a distinct human voice with rhetorical questions, personal hedging ('My guess...'), colloquialisms ('butt-load') and emotional appeals that are typical of forum-style human writing. Second, there are multiple typographical and factual inconsistencies (missing space in 'Wahhabi... | {
"rationale": "First, the passage shows a distinct human voice with emotional appeals, rhetorical questions, colloquialisms ('butt-load'), and personal bets ('dollars to doughnuts') suggesting spontaneous human expression. Second, there are typographic and factual inconsistencies (missing space after a period, uneve... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 1,251 |
human-1743 | human | Having to worry about medical bills should not be a problem for anyone. People in the States are so worried about big government, but I strongly believe we need a restructuring of our health care program. 2 years ago I had about 15,000 in student loan debt, with no other debts to my name. Last year I was involved in an... | NA | cmv | {
"rationale": "First, the post uses forum-specific shorthand ('CMV') and personal, anecdotal details (specific loan amount, accident) that match human social-media behavior. Second, there are casual stylistic choices, inconsistent capitalization, and a minor grammatical slip ('I've' vs 'I have') suggesting spontaneo... | {
"rationale": "First, the post uses forum-specific shorthand ('CMV') and personal, anecdotal details (specific debt amount, accident) which are typical of human social-media posts. Second, there are small stylistic inconsistencies and colloquial phrasings ('I strongly believe', 'than I would of elsewhere') suggestin... | {
"rationale": "First, the post uses forum-specific shorthand ('CMV') and personal, anecdotal details (specific debt amount, accident) which are typical of human social-media posts. Second, there are small stylistic inconsistencies and colloquial phrasings ('I strongly believe', 'than I would of elsewhere') suggestin... | {
"rationale": "First, the post uses forum-specific shorthand ('CMV') and personal, anecdotal details (specific debt amount, accident) which are typical of human social-media posts. Second, there are colloquial errors and inconsistencies (missing hyphen in 'I'd of', mixed capitalization, uneven spacing) suggesting sp... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 266 |
human-299 | human | We have struggled for ages to deal with issues of police abuse of power and force, and it seems like we're only beginning to scratch the surface. With many police departments recently (in the last 5 years) under investigation by the Federal DoJ, it's readily apparent that this is not a one-off problem, it's a systemic ... | NA | cmv | {
"rationale": "First, the post shows a conversational, forum-style voice with personal hedging ('I think', 'I'm not really sure') and an explicit edit note, patterns common in human online posts. Second, it contains typographical errors and idiosyncratic phrasings ('theythat', 'powerforce') and casual asides ('EDIT.... | {
"rationale": "First, the post has a highly conversational, forum-style voice with personal hedging ('I think', 'I'm not really sure') and an explicit 'EDIT:' marker typical of human online posts. Second, it shows idiosyncratic errors and informal choices (missing space in 'themthat', casual phrasings, specific loca... | {
"rationale": "First, the post shows a conversational, forum-style voice with personal hedging ('I think', 'I'm not really sure') and an explicit edit note, patterns common in human online posts. Second, there are typographical errors and idiosyncratic phrasings ('theythat', 'powerforce') suggesting casual human typ... | {
"rationale": "First, the passage shows a highly conversational, forum-style voice with personal hedging ('I think', 'I'm not really sure') and an explicit edit note, suggesting spontaneous human posting. Second, it contains typographical errors and idiosyncratic phrasings ('theythat', 'powerforce') and uneven struc... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 444 |
human-895 | human | [Research has shown] that Squirtle is more effective against gym leaders and the elite four, and has less not very effective tendencies against those same trainers than the other two starters. Squirtle and its evolved forms can learn both swim and cut where as Bulbasaur can only learn cut. Then when it comes to the 'co... | NA | cmv | {
"rationale": "First, the post uses a casual forum voice (e.g., 'Please Change My View') and includes personal markers like 'I' and 'I will be', which are typical of human online posts. Second, there are small typos and inconsistencies ('where as' for 'as' and mixed capitalization of The Squirtle Squad) suggesting s... | {
"rationale": "First, the post uses a casual forum voice (e.g., 'Please Change My View') and includes community-specific cues like bracketed research notes and references to other players, which is typical of human online posts. Second, there are small typos and inconsistencies ('where as' for 'as' and mixed capital... | {
"rationale": "First, the post uses a casual forum tone, personal stakes ('I will be open-minded...'), and a direct plea ('Please Change My View') common in human online posts. Second, there are small typos and inconsistencies (e.g., 'where as' instead of 'as' and mixed capitalization) suggesting spontaneous human t... | {
"rationale": "First, the post uses a casual forum tone, personal stakes ('I will be open-minded...'), and a direct plea ('Please Change My View') common in human online posts. Second, there are small typos and inconsistencies (e.g., 'where as' instead of 'as' and mixed capitalization) suggesting spontaneous human t... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 247 |
human-1978 | human | before you guys bombard me with downvotes i would simply like to point out that i am not saying this is true. i am just expressing my conditional views that i am forced to live with. these are obviously the views i wish you guys could help me conquer. to get started i'm pretty sure that by now you guys are probably thi... | NA | cmv | {
"rationale": "First, the passage contains highly idiosyncratic, emotionally raw first-person details, slang, and forum-specific markers (e.g., 'rtheredpill', explicit CMV moderator footnote) consistent with copied human Reddit text. Second, it shows non-linear thought patterns, inconsistent grammar, and colloquial ... | {
"rationale": "First, the passage contains highly contextual, community-specific details (rtheredpill, CMV moderator footnote) and colloquial phrasings that match human forum posts. Second, it shows inconsistent grammar, typos, profanity, and emotionally raw, self-reflective narration suggesting spontaneous human ty... | {
"rationale": "First, the passage contains highly contextual, community-specific details (rtheredpill, CMV moderator footnote) and colloquial, emotionally charged phrasing typical of human forum posts. Second, it shows numerous typos, inconsistent grammar, and idiosyncratic repetition ('i been stuck with') suggestin... | {
"rationale": "First, the passage contains highly idiosyncratic, emotionally raw, and context-specific details (rtheredpill, CMV moderator footnote) consistent with copied human forum text. Second, it shows numerous casual colloquialisms, spelling/grammar slips ('they only want a penis that...'), inconsistent punctu... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 668 |
human-937 | human | In my state of Georgia, the state government put in place a total ban on undocumented students from attending the top 5 universities in the University System of Georgia. For all other schools in the state, they raised the price of tuition for undocumented students to the price of out-of-state tuition, making it financi... | NA | cmv | {
"rationale": "First, the passage contains personal anecdotes, emotional nuance, and specific local details (Georgia, University System of Georgia) that point to a real human author. Second, it shows typographic and grammatical slips ('Ive', 'struggled' for 'strated') and uneven sentence flow typical of spontaneous ... | {
"rationale": "First, the passage contains personal anecdotes, emotional engagement (e.g., 'The undocumented youth...'), and specific local details (Georgia, University System of Georgia) that align with human-authored opinion writing. Second, there are small stylistic inconsistencies and typographic slips ('Ive' wi... | {
"rationale": "First, the passage contains personal anecdotes, emotional engagement (e.g., 'The undocumented youth...'), and specific local details (Georgia, University System of Georgia) suggesting a human author. Second, there are typographic and stylistic idiosyncrasies ('Ive', uneven pacing, rhetorical questions... | {
"rationale": "First, the passage contains personal anecdotes, emotional engagement, and specific local details (Georgia, University System of Georgia) suggesting a human author. Second, there are typographic and grammatical idiosyncrasies ('Ive', uneven pacing, long descriptive sentence) more typical of spontaneous... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 314 |
human-1869 | human | Hi CMV! I'm out to genuinely seek to understand this perspective, because I confess it rubs me the wrong way when I hear it discussed in media, here on reddit or anywhere - "America is the land of the free" is a slogan we all know and lovehate in this world. I believe that it is clearly propaganda, as is all discussion... | NA | cmv | {
"rationale": "First, the post shows a distinct conversational Reddit voice with typos, bracketed moderator footnotes, and community-specific markers ('CMV', 'Happy CMVing!') that are typical of human forum posts. Second, it contains idiosyncratic phrasings, hedging, personal anecdotes (UK age difference, local drin... | {
"rationale": "First, the post shows forum-specific markers (CMV, moderator footnote, bracketed link placeholders) and a conversational, anecdotal voice typical of human Reddit posts. Second, it contains idiosyncratic typos, inconsistent spacing, hedging language ('I believe', 'I feel'), and uneven structure suggest... | {
"rationale": "First, the post shows forum-specific markers (CMV, moderator footnote, bracketed link placeholders) and a conversational, anecdotal voice typical of human Reddit posts. Second, it contains idiosyncratic typos, inconsistent spacing, hedging language ('I believe', 'I feel'), and uneven structure suggest... | {
"rationale": "First, the post shows a highly conversational, forum-specific voice with typos, bracketed link placeholders, and moderator footer text consistent with a copied Reddit CMV thread rather than polished AI output. Second, it contains idiosyncratic phrasings, personal anecdotes ('apologies, it's what I kno... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 1,376 |
human-1441 | human | First off, a bit of background on me so you know where I'm coming from. I was in the class of '05 and my highschool had a deal with Coke; it was on our scoreboards, our lunchroom had a big Coke clock, and we had 3 Drink Machines (2 Coke, 1 Powerade) and a food vending machine with candy. And to say I partook would be a... | NA | cmv | {
"rationale": "First, the passage contains highly specific personal details (class of '05, Mello Yello, exact consumption patterns) and a conversational, anecdotal voice typical of human forum posts. Second, there are small typographic and stylistic idiosyncrasies ('weight managementobesity', 'MorbidClass II Obesity... | {
"rationale": "First, the post contains highly specific personal details (class of '05, Mello Yello, exact consumption patterns) and a conversational, anecdotal tone typical of human forum posts. Second, there are small typographic and stylistic inconsistencies (missing space in 'weight managementobesity', mixed hyp... | {
"rationale": "First, the post contains highly specific personal details (class of '05, exact beverage intake, medical history) and a conversational, anecdotal tone typical of human forum posts. Second, there are small typographic and stylistic inconsistencies (missing space in 'weight managementobesity', varied sen... | {
"rationale": "First, the passage contains highly specific personal details (class of '05, Mello Yello, exact consumption patterns) and a conversational, anecdotal tone typical of human forum posts. Second, there are typographic and formatting idiosyncrasies (missing space in 'weight managementobesity', mixed hyphen... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 605 |
human-50 | human | DISCLAIMER: I am currently studying economics and mandarin, so it's safe to say I'm not a "DAE le STEM xD" guy. 1) Subjects such as genderspecific culture studies, economics and sociology etc. are completely ideological. Your academic worth is decided by how closely your opinions reflect the approved view. This manifes... | NA | cmv | {
"rationale": "First, the post shows a distinct personal voice, forum-specific markers (TL;DR, CMV), and context-dependent references like a moderator footnote that match human online posting. Second, it contains idiosyncratic typos and formatting oddities (e.g., 'genderspecific', 'andor') and emotionally charged, i... | {
"rationale": "First, the post shows forum-specific markers (TL;DR, CMV moderator footnote, bracketed link placeholders) and a personal, idiosyncratic voice consistent with human online posting. Second, there are typographical errors and informal inconsistencies (e.g., 'genderspecific', 'andor') plus emotionally cha... | {
"rationale": "First, the post shows forum-specific markers (TL;DR, CMV moderator footnote, bracketed link placeholders) and a conversational tone typical of human Reddit posts. Second, there are idiosyncratic errors and informal phrasings ('genderspecific', 'andor', 'studiying') and inconsistent spacing/parentheses... | {
"rationale": "First, the post displays a highly conversational, idiosyncratic voice with personal anecdotes, slang (e.g., 'andor'), and community-specific markers ('TL;DR', CMV) consistent with human forum posts. Second, there are typographical errors, inconsistent spacing/parentheses, and copy-paste artifacts (mod... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 936 |
human-685 | human | Pretty simple, really. George Lucas is an extraordinary man, and his contributions to cinema are pretty much undeniable. That being said, the OT trilogy is not, or it rather shouldn't be. For one very simple reason: Lucas is an idea man. As evidenced by the prequel trilogy, he has some ideas ranging from decent to outr... | NA | cmv | {
"rationale": "First, the post uses a highly conversational, forum-style voice with personal markers ('Pretty simple, really', 'CMV!') and emphatic punctuation/ellipses that match human online writing. Second, it shows idiosyncratic errors and stylistic inconsistencies—misspellings like 'changefix', inconsistent cap... | {
"rationale": "First, the post uses highly conversational, forum-style markers ('Pretty simple, really', 'CMV!', multiple exclamation marks) and personal voice typical of human online posts. Second, it shows idiosyncratic errors and stylistic inconsistencies (run-on sentences, misspelling 'changefix', ALL CAPS empha... | {
"rationale": "First, the post uses highly conversational, forum-specific markers ('Pretty simple, really', 'CMV!', multiple exclamation marks) and colloquial phrasings common in human online replies. Second, it shows personal voice, emotional engagement, inconsistent punctuation/spacing and small stylistic slips (e... | {
"rationale": "First, the post uses highly conversational, forum-style phrasing ('Pretty simple, really', 'CMV!') and personal voice consistent with human online comments. Second, it shows idiosyncratic errors and stylistic quirks—typos ('changefix'), inconsistent capitalization, excessive exclamation and repetition... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 544 |
human-1608 | human | To make it clear from the beginning: I am not talking about legitimate rape cases, nor am I implying that every time a male is found not guilty of rape, the female should go to prison. I am talking about persons that abuse the system and make blatantly false allegations to the police having an ulterior motive in mind. ... | NA | cmv | {
"rationale": "First, the post shows a distinct conversational voice with personal touches, slang, emoticon use, and an 'Edit' note, patterns common in human forum posts. Second, it contains casual typos, inconsistent punctuation, bracketed asides, and idiosyncratic phrasings ('whatshisname', 'fire away') suggesting... | {
"rationale": "First, the post shows a highly conversational, forum-style voice with personal touches (bracketed link placeholder, 'Edit: damn formatting', emoticon) typical of human online posts. Second, it contains idiosyncratic errors and colloquialisms ('whatshisname', inconsistent spacing, run-on explanations) ... | {
"rationale": "First, the post shows a highly conversational, forum-style voice with personal touches (bracketed link placeholder, 'Edit: damn formatting'), typos ('whatshisname') and emoticons, patterns common in human online posts. Second, it contains inconsistent punctuation, colloquial slang ('disgruntled over-e... | {
"rationale": "First, the post shows a highly conversational, forum-style voice with personal touches (bracketed link placeholder, 'Edit: damn formatting'), typos ('whatshisname') and emoticons, patterns common in human online posts. Second, it contains inconsistent punctuation, emotional hedging, and idiosyncratic ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 671 |
human-396 | human | So here are some main points: The metals, plastics, and wood fibers temporarily being stored in our landfills are valuable and as resource scarcity increases, the value of these materials will continue to increase. One of the main hurdles preventing the mining of landfills and wide-scale, profitable recycling today is ... | NA | cmv | {
"rationale": "First, the passage has a distinct forum-style voice with community shorthand ('CMV') and personal asides ('Anyone...'), which is typical of human online posts. Second, it uses rhetorical questions, parenthetical uncertainty ('(billions?)'), concrete cultural references (Halliburton, abject poverty) an... | {
"rationale": "First, the post has a conversational, forum-style voice with community markers ('CMV', 'Anyone...') and personal asides that match human social-media posting. Second, it uses rhetorical questions, concrete examples (Halliburton, abject poverty) and mild emotional bias ('being spent by people that have... | {
"rationale": "First, the post has a conversational, forum-style voice with community markers ('CMV', 'Anyone...') and personal asides that match human social-media posting. Second, it shows idiosyncratic phrasings, rhetorical questions, parentheses with uncertainty ('(billions?)'), and concrete cultural references ... | {
"rationale": "First, the post has a highly conversational, forum-style voice with community-specific markers ('CMV', 'abject poverty') suggesting a human Reddit poster. Second, it contains idiosyncratic phrasings, rhetorical questions, parentheses with uncertainty ('(billions?)'), and concrete personal references (... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 453 |
human-1890 | human | APOLOGIES FOR THE STUPID TITLE! This is the summary of my view: I believe that consenting to sex is not consenting to reproduction and that it is unethical not to have an abortionto expect your partner not to abort in case of an accident. I believe that using contraception andor discussing this in advance is equal to t... | NA | cmv | {
"rationale": "First, the post shows a highly personal, idiosyncratic voice with context-specific references (reddit, 'I am female') and nuanced emotional details that are typical of human-authored essays. Second, it contains typographical errors, inconsistent spacing, colloquial asides, and self-edits ('EDIT...') w... | {
"rationale": "First, the post shows a highly conversational, personally detailed Reddit-style voice with context-specific references (e.g., 'on reddit') and an 'EDIT' note, which is typical of human forum posts. Second, it contains idiosyncratic typos and inconsistencies ('abortionto', 'andor', 'in fact I surprised... | {
"rationale": "First, the post shows a highly conversational, personally detailed Reddit-style voice with anecdotes, emotional reactions, and explicit edits ('EDIT...'), which is typical of human forum posts. Second, it contains idiosyncratic typos and inconsistencies (e.g., 'abortionto', 'andor') and nuanced moral ... | {
"rationale": "First, the post shows a highly personal, idiosyncratic voice with emotional nuance, anecdotes, and forum-specific markers (apology for title, reddit inspiration, EDIT) that are typical of human-authored online posts. Second, it contains typographical errors, inconsistent spacing, colloquial phrasings,... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 1,789 |
human-1716 | human | I've heard parents say that it is to hard to prevent your kids from running away. But personally I think leashing is really silly. You can simply hold hands like most parents do. You are limiting your kids ability to discover. If Little Timmy wants to go run to the bey blade section at the store, chase after him but le... | NA | cmv | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, anecdotal tone typical of human online posts. Second, it shows idiosyncratic errors and colloquialisms ('it is to hard', 'there safety') and emotional shifts that AIs rarely produce so n... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and conversational, self-correcting tone typical of human online posts. Second, it shows casual errors and idiosyncrasies—'it is to hard', 'its a bad thing', inconsistent capitalization—that AIs rarely produc... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and conversational, self-correcting tone typical of human online posts. Second, it shows casual errors and idiosyncrasies ('it is to hard', 'its a bad thing') suggesting spontaneous human writing rather than ... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and conversational, self-correcting tone typical of human online posts. Second, it shows idiosyncratic errors and colloquialisms ('it is to hard', 'there safety') suggesting spontaneous human writing rather t... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 442 |
human-1571 | human | Not for any religious reasons, but for the sake of human biology and science. Firstly, let me give some of my background. I did not grow up in a religious home, and I've always had a pretty liberal outlook on society. I grew up around many gay people on the outskirts of San Francisco and never had any problem with homo... | NA | cmv | {
"rationale": "First, the passage contains personal anecdotes, emotional nuance, and idiosyncratic phrasings (e.g., 'not for any religious reasons... for the sake of human biology and science') that match human conversational style. Second, it shows inconsistent argumentation, rhetorical risk-taking, and specific li... | {
"rationale": "First, the passage has a highly personal, anecdotal voice with specific lived details (San Francisco, Tumblr, a video of a ' bald lesbian') and emotional nuance that reads like human testimony. Second, it shows inconsistent argumentation, conversational hedges ('Try to change my view'), and moral ambi... | {
"rationale": "First, the passage shows a distinct personal voice, specific lived details (San Francisco, Tumblr, a friend) and emotional nuance that are typical of human-authored essays. Second, it contains idiosyncratic phrasings, small inconsistencies and moral ambivalence ('not grow up in a religious home' then ... | {
"rationale": "First, the passage contains highly personal, emotionally nuanced details (specific social experiences, named sites like Tumblr) and a conversational, idiosyncratic voice typical of human-authored posts. Second, it shows uneven argumentation, repetition, and occasional logical slips rather than the pol... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 539 |
human-1954 | human | I have had this discussion many times on reddit. And there are many good points to be made for piracy. However, to me, it boils down to this fundamental point: That the creator of something has the fundamental right to do whatever he chooses with it. If the creator of a video gamemoviesongbook chooses to sell it, and y... | NA | cmv | {
"rationale": "First, the passage contains forum-specific cues (mentions Reddit) and a conversational, anecdotal tone with typos and inconsistent spacing ('how' for 'hop', 'with out') typical of human casual posts. Second, it shows personal experience, emotional appeals ('Please, change my view'), and culturally spe... | {
"rationale": "First, the post shows colloquial, forum-specific phrasing (e.g., 'I have had this discussion many times on reddit') and personal appeals ('Please, change my view'), which are typical of human online posts. Second, it contains typographical errors and inconsistent spacing ('how' for 'hub', 'gamemovieso... | {
"rationale": "First, the post shows informal, forum-specific phrasing (e.g., 'I have had this discussion many times on reddit') and personal appeals ('Please, change my view'), patterns common in human online posts. Second, it contains typographical errors and idiosyncratic token choices ('gamemoviesongbook', missi... | {
"rationale": "First, the passage shows a highly conversational, forum-specific voice with personal anecdotes (\"I have had this discussion many times on reddit\") and Reddit-style phrasing (\"Please, change my view\"), which is typical of human posts. Second, it contains casual typos and inconsistencies (e.g., \"ga... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 372 |
human-2128 | human | I feel the inequality in wealth in most nations is terrible, however, I think a general wage cap would be horrible to our economy and kill incentive. I think a novel approach would be to tie a wage cap to the average income. For example make the maximum wage ten thousand times the average income of everyone else. If th... | NA | cmv | {
"rationale": "First, the passage shows conversational, idiosyncratic phrasing and personal markers (e.g., 'I' opening, rhetorical question 'Am I completely wrong?') typical of human forum posts. Second, there are small stylistic inconsistencies and informal choices ('however,' abrupt transitions, casual numeric spe... | {
"rationale": "First, the passage shows conversational, forum-style phrasing (e.g., 'I feel...','Am I completely wrong?') and personal hedging typical of human posts. Second, there are small stylistic inconsistencies and colloquialisms ('however,' abrupt transitions, '1)') suggesting spontaneous human writing rather... | {
"rationale": "First, the passage shows conversational, forum-style phrasing and personal uncertainty (e.g., 'Am I completely wrong?') typical of human posts. Second, there are small stylistic inconsistencies and colloquialisms ('however,' abrupt transitions, numeric exaggeration) suggesting spontaneous human typing... | {
"rationale": "First, the passage shows a conversational, forum-style tone with personal hedging ('I feel', 'Am I completely wrong?') typical of human posts. Second, there are small stylistic inconsistencies (e.g., 'however' followed by 'I think') and casual phrasings ('in most nations') suggesting spontaneous human... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 288 |
human-1884 | human | I've watched every single episode of Community that has been aired and I think that for the most part every season is more or less equal. I've heard nothing but horrible things from critics and fans alike about Season 4, but didn't really notice any obvious differences in how the characters interacted, the quality of t... | NA | cmv | {
"rationale": "First, the passage shows a conversational, idiosyncratic voice with personal anecdotes and emotional investment (e.g., 'I've watched...') typical of human forum posts. Second, it contains small stylistic inconsistencies, colloquial phrasings ('didn't really notice'), and a minor typo/spacing issue ('t... | {
"rationale": "First, the passage shows a conversational, idiosyncratic voice with personal anecdotes and emotional investment (e.g., 'I've watched...') typical of human forum posts. Second, there are small stylistic inconsistencies and colloquialisms ('didn't really notice', 'created by people looking...') suggesti... | {
"rationale": "First, the passage shows a conversational, idiosyncratic voice with personal anecdotes and emotional investment (e.g., 'I've watched...') typical of human forum posts. Second, there are small stylistic inconsistencies and colloquialisms ('didn't really notice... because of their devotion...'), suggest... | {
"rationale": "First, the passage shows a highly personal, conversational voice with colloquial phrasing (I've watched...; I truly don't feel like) typical of human forum posts. Second, there are small stylistic inconsistencies and idiosyncrasies—repetition, casual tone, and the odd word choice tonetheme—suggesting ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 289 |
human-210 | human | I work in an contemporary art gallery so this might be a bit odd but ya I think most of the work our partners do is fucking stupid. Example: One of our partners is working with an artist called Robert Foster and he "performs" his art. His most recent piece had him standing with a old sheet over him for a few hours. The... | NA | cmv | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a messy, conversational tone consistent with human online posting. Second, there are typographical errors ('an contemporary') and colloquialisms ('ya') plus uneven spacing and abrupt transitions that AI r... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a messy, conversational tone typical of human online posts. Second, there are typographical errors ('an contemporary') and colloquial phrasings ('ya') suggesting spontaneous human typing rather than polis... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a messy, conversational tone consistent with human copy-paste. Second, there are typographical errors ('an contemporary') and colloquialisms ('ya') suggesting casual human typing rather than polished AI o... | {
"rationale": "First, the passage contains highly contextual, platform-specific elements (CMV, moderator footnote, bracketed links) consistent with a copied Reddit post rather than generic AI output. Second, there are casual colloquialisms ('ya'), spelling/grammar slips ('an contemporary', 'a old') and uneven phrasi... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 385 |
human-1553 | human | I come from a poor background in a very left wing country (the Netherlands) myself, and so was raised with the idea that all people should be taken care of in a relatively equal manner, but I now consider such a situation disastrous and unsustainable. I mention this, because I don't want people to assume that I'm merel... | NA | cmv | {
"rationale": "First, the passage shows a distinct personal voice, cultural references, and a satirical, emotionally charged tone that reads like an individual's creative essay rather than a neutral AI summary. Second, it contains idiosyncratic errors and stylistic quirks (e.g., 'gunpoint', bracketed 'some', uneven ... | {
"rationale": "First, the passage shows a distinct personal voice, cultural references (e.g., 'the Netherlands'), and a satirical, argumentative tone typical of human-authored posts. Second, it contains idiosyncratic errors and stylistic quirks ('gunpoint', bracketed 'some', uneven transitions) suggesting spontaneou... | {
"rationale": "First, the passage shows a distinct personal voice, cultural specificity (the Netherlands), and emotionally charged rhetorical flourishes that are typical of human-authored opinion writing. Second, it contains idiosyncratic errors and stylistic inconsistencies (e.g., 'gunpoint', bracketed 'some', unev... | {
"rationale": "First, the passage shows a distinct personal voice, vivid cultural references (Twilight, Fifty Shades), and a satirical, emotionally charged tone typical of human online posts. Second, it contains idiosyncratic errors and stylistic inconsistencies (e.g., 'gunpoint', bracketed 'some', uneven transition... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 1,092 |
human-1561 | human | I am attending my country's top school for math and computer science. I just feel like my time would be best spent elsewhere. The things I am learning are pretty advanced, but I dont think I am learning anything useful. All we do is come up with toy models to show how some complex theory is used. I feel as if my time w... | NA | cmv | {
"rationale": "First, the passage contains highly specific personal details, colloquial phrasing, and emotional nuance (e.g., 'I guess one could say...') that match genuine human forum posts. Second, it includes contextual artifacts — a Reddit CMV moderator footnote with bracketed links and community-specific cues —... | {
"rationale": "First, the post shows colloquial, context-specific details (CMV subreddit, moderator footnote, exact habits) and small inconsistencies typical of human forum posts. Second, there are typos ('dont') and uneven phrasing ('I guess one could say...') suggesting spontaneous human writing rather than polish... | {
"rationale": "First, the post shows highly specific personal details, colloquial tone, and emotional nuance (e.g., explicit profanity and sleep/time references) typical of human forum posts. Second, there are small inconsistencies and stylistic idiosyncrasies—like missing apostrophes, casual phrasing, and a pasted ... | {
"rationale": "First, the passage contains highly specific personal details (attendance, sleep times, coursework) and colloquial phrasing ('give a shit') consistent with a real Reddit post rather than generic AI output. Second, it includes forum-specific artifacts (CMV footer, bracketed links, moderator note) sugges... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 552 |
human-203 | human | I have worked with, befriended, and dated feminists for many years. I'm not a red-piller (and indeed am shocked by many of their beliefs), but after a lot of debate I have come to the parallel conclusion that the feminist movement is a sexist hate group that attacks both men and masculinity: A team-based [us vs. them m... | NA | cmv | {
"rationale": "First, the passage contains idiosyncratic typos, bracketed edits, and a distinct personal voice (e.g., 'I'd to head off') that are typical of human forum posts. Second, it uses emotionally charged, context-specific examples ('SFW', catcalling, raping or murdering our wives) and rhetorical flourishes s... | {
"rationale": "First, the passage has a distinct personal voice, colloquial phrasing, and rhetorical flourishes (e.g., 'I'd to head off...') that are typical of human forum posts. Second, it contains typographical errors, inconsistent spacing/bracket use, and idiosyncratic word choices ('andor', bracket misuse) sugg... | {
"rationale": "First, the passage shows a distinct personal voice, colloquial phrasing, and rhetorical flourishes (e.g., 'I'd to head off...') that are typical of human forum posts. Second, there are typographic idiosyncrasies, inconsistent spacing/bracket use, and emotionally charged hyperbole ('sexist hate group')... | {
"rationale": "First, the passage shows a distinct personal voice, colloquial phrasing, and numerous typographical/spacing errors (e.g., 'andor', missing brackets) consistent with copied human forum text. Second, it contains emotionally charged, idiosyncratic examples (' raping or murdering our wives') and rhetorica... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 822 |
human-891 | human | Reasons why I think it is bad: -it is unrealistic -there are many events that are extraneous to the plot (i.e. the main character gets laid for no reason every episode) -there is very little character development (I've seen the first season) -everything seems kind of '90searly 2000's. This could be good or bad, but it ... | NA | cmv | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational, self-correcting Reddit-style voice consistent with human posting. Second, it shows idiosyncratic errors and informal choices ('90searly', 'whore-ish') and uneven grammar/typos typical of... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational Reddit-style voice consistent with human posting. Second, it shows idiosyncratic typos and informal phrasings ('90searly', 'whore-ish') and uneven structure that humans commonly produce b... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational Reddit-style voice consistent with human posting. Second, it shows idiosyncratic typos and informal phrasings ('90searly', 'whore-ish') and uneven editing that humans commonly produce but... | {
"rationale": "First, the passage contains forum-specific markers (CMV, moderator footnote, bracketed links) and a conversational Reddit-style voice consistent with human posting. Second, it shows idiosyncratic typos and informal phrasings ('90searly', 'whore-ish') and uneven editing suggesting spontaneous human typ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 599 |
human-994 | human | I believe that people who say that they don't want to live forever are just comforting themselves with fake reasons to accept the inevitability of death. As a result of this we have too much acceptance of death and spend too little effort on life extension and eventual immortality. Some clarifications below. 1. By immo... | NA | cmv | {
"rationale": "First, the passage contains forum-specific markers (UPD, bracketed link, '- - -') and a conversational, idiosyncratic voice typical of human online posts. Second, it shows casual typos and inconsistent spacing ('Remeber') plus personal asides ('I will reformat...') suggesting spontaneous human composi... | {
"rationale": "First, the passage shows forum-specific markers (UPD, bracketed link placeholder, '- - -') and a conversational, stream-of-consciousness tone typical of human posts. Second, there are idiosyncratic errors and informal choices ('Remeber', '800 or 8000') and blunt profanity ('bullshit') suggesting spont... | {
"rationale": "First, the passage shows forum-specific markers (UPD, bracketed link placeholder, '- - -') and a conversational, stream-of-consciousness tone typical of human posts. Second, there are idiosyncratic errors and informal choices ('Remeber', '800 or 8000') and blunt profanity ('bullshit') suggesting spont... | {
"rationale": "First, the passage shows forum-specific markers (UPD, bracketed link placeholder, '- - -') and a conversational, messy structure typical of human posts. Second, it contains idiosyncratic phrasings, profanity, and personal asides ('I will reformat...') suggesting spontaneous human writing rather than p... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 491 |
human-150 | human | Fox News defends Republican politicians and bash liberal politicians. Other news networks are vice versa. Imagine if we had a mainstream media that defended regular people and the community rather than focus on the capitalist profit. Example. Mainstream media will talk about "The next move for Snowden" rather than talk... | NA | cmv | {
"rationale": "First, the post uses casual forum markers (CMV), contractions, and specific cultural references (Fox News, Zimmerman trial) which are common in human online posts. Second, it shows uneven phrasing, personal voice, and minor stylistic slips ('bash liberal politicians' instead of 'labeling') suggesting ... | {
"rationale": "First, the post uses casual forum markers (CMV), contractions, and a personal, opinionated voice typical of human social-media posts. Second, it shows uneven phrasing, repetition ('Other...'), and small stylistic slips ('versus' for 'vice versa') suggesting spontaneous human writing rather than polish... | {
"rationale": "First, the post uses casual forum markers (CMV), slang ('bash'), and uneven sentence structure typical of quick human posts. Second, there are small inconsistencies and stylistic slips ('vice versa' for networks) suggesting spontaneous human writing rather than polished AI output. Moreover, it present... | {
"rationale": "First, the post uses casual forum markers (CMV), slang ('bash'), and uneven sentence structure typical of quick human social-media posts. Second, it contains personal viewpoint, emotional stance, and specific cultural references (Fox News, Zimmerman trial) suggesting lived experience rather than gener... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 244 |
human-1172 | human | It seems that every time part of the Arab world is given the chance to elect their own government, they always bollocks it up. I think it's the reason there are so many authoritarian regimes in the Middle-East - oppressive dictatorships and autocracies that keep it's citizens on a very short leash are the only ways to ... | NA | cmv | {
"rationale": "First, the passage shows colloquial phrasing, contractions ('bollocks'), and a spelling/grammar slip ('it's citizens') that are typical of quick human typing. Second, it uses personal markers ('I think', 'I'm') and contains uneven sentence lengths and stylistic idiosyncrasies suggesting individual voi... | {
"rationale": "First, the passage uses colloquial phrasing, contractions (e.g., 'bollocks') and a personal voice ('I think'), which are common in human forum posts. Second, it shows minor grammatical slips ('it's citizens' from devolving) and uneven sentence structure suggesting spontaneous typing rather than polish... | {
"rationale": "First, the passage shows colloquial phrasing, spelling/grammar slips (e.g., 'it's citizens') and blunt profanity ('bollocks'), patterns common in spontaneous human writing. Second, it uses personal markers ('I think', 'I'm') and offers tentative, opinionated reasoning rather than the more polished, ne... | {
"rationale": "First, the passage shows colloquial, opinionated phrasing and multiple typos/grammatical slips (e.g., 'bollocks' for 'bulls' and misuse of 'it's') more typical of quick human typing than polished AI output. Second, it contains personal voice and hedging ('I think...'), emotional stance, and uneven sen... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 255 |
human-2351 | human | This is influenced by the IAmA submission from the guy who proudly kills fellow human beings in the name of "saving" wildlife. But really? How are poachers more morally and ecologically irresponsible than hunters? Hunters have been responsible for the extinction of wildlife in recent memory and in past events of human ... | NA | cmv | {
"rationale": "First, the passage shows a distinct personal voice, colloquial phrasing (e.g., 'the guy... in the name of ... saving wildlife') and rhetorical questions that are typical of human online posts. Second, it contains specific cultural references ('IAmA'), emotional stance, and inconsistent sentence-level ... | {
"rationale": "First, the post uses a casual, conversational tone with forum-specific shorthand ('IAmA') and a personal anecdote, which matches human online behavior. Second, it shows idiosyncratic phrasing and minor grammatical slips (e.g., 'in the name of ...' missing 'the') and emotional emphasis ('But really?'),... | {
"rationale": "First, the post uses a casual, conversational tone with forum-specific shorthand ('IAmA') and slang ('guys'), suggesting a human online voice. Second, it shows idiosyncratic emphasis (all-caps 'Proudly kills...') and abrupt transitions typical of spontaneous human writing rather than polished AI outpu... | {
"rationale": "First, the post displays a highly conversational, emotionally charged tone with forum-specific shorthand ('IAmA') and casual phrasing ('but really?'), suggesting spontaneous human posting. Second, it contains idiosyncratic errors and colloquialisms (e.g., 'in wildlife' missing 'the' and inconsistent c... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 269 |
human-2008 | human | First off, I come from the US, where drinking and driving are spoken against but not acted against. I don't think it should be a long sentence, no need for life terms or anything. However, I do think more drastic steps should be taken to stop persons from consuming and driving. If one were to look at the harshness of d... | NA | cmv | {
"rationale": "First, the passage shows a conversational, idiosyncratic voice with colloquial phrasings ('no need for life terms or anything', 'what do you all think?') typical of human forum posts. Second, it contains small stylistic inconsistencies and local cultural references (US, Europe, 'drunk driving') sugges... | {
"rationale": "First, the passage has a casual, conversational tone with personal details (US, Europe) and colloquial phrasings ('what do you all think') typical of human forum posts. Second, it shows small stylistic inconsistencies and redundancy ('no need for life terms or anything', 'injunction with the rate of o... | {
"rationale": "First, the passage shows a casual, conversational tone with colloquial phrasings ('what do you all think?', 'no need for life terms or anything') typical of human forum posts. Second, there are small stylistic inconsistencies (e.g., 'in junction' vs 'in relation to', missing hyphen in 'drinking and dr... | {
"rationale": "First, the passage shows a casual, conversational tone with personal details (\"First off...from the US\") and colloquial phrasing (\"do you all think?\") typical of human forum posts. Second, it contains small stylistic inconsistencies and redundancy (e.g., \"in junction with\" vs \"in relation to\",... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 285 |
gpt-3.5-turbo-69 | ai | There's a federal law which prohibits anyone from buying a vote or paying someone to withold their vote, or accepting anything for a vote. The spirit of this law is to ensure that elections are fair and that voters are not unduly influenced by bribery or other forms of corruption. It is designed to promote transparency... | NA | cmv | {
"rationale": "First, the passage uses polished, generic explanatory language and frequent transitional phrases (e.g., 'essentially', 'it is designed') that are common in AI-generated text. Second, it lacks concrete specifics, personal nuance, or contextual grounding and instead offers high-level generalizations, su... | {
"rationale": "First, the passage uses polished, generic explanatory language and repeats key phrases (e.g., 'essentially', 'spirit') in a way common to AI-generated text. Second, it lacks concrete specifics, examples, or personal nuance about laws or processes and instead offers high-level generalizations, suggesti... | {
"rationale": "First, the passage uses polished, generic explanatory language and lacks concrete specifics (no statute or case cited), a common trait of AI-generated text. Second, it repeats ideas with smooth transitions and balanced sentence lengths, suggesting algorithmic uniformity rather than human stylistic var... | {
"rationale": "First, the passage uses very polished, generic explanatory language and lacks concrete details (no specific law name or examples), a common trait of AI-generated text. Second, it repeats ideas in balanced, symmetric clauses ('essentially... essentially') suggesting algorithmic paraphrasing rather than... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 269 |
gpt-3.5-turbo-51 | ai | [This article] does a pretty good job of pulling the relevant statistics together that I find convincing. In searching for critiques of those statistics I found [this article] that lays out some alternative perspectives and criticisms of the original statistics presented in the first article. It's important to consider... | NA | cmv | {
"rationale": "First, the passage uses generic, balanced phrasing and lacks concrete details or examples, which is common in AI-generated text. Second, its structure is formulaic—intro, neutral summary, brief rebuttal, and a closing generalization—and it repeats transitional phrases suggesting template-like generati... | {
"rationale": "First, the passage uses generic, balanced phrasing ('does a pretty good job... I find convincing' and 'It's important to consider...') which is common in AI-generated text. Second, it lacks concrete details (no authors, articles, or specific statistics) and repeats simple structures ('[This article] .... | {
"rationale": "First, the tone is generic and evenly balanced with stock phrases ('does a pretty good job...I find convincing', 'It's important to consider...'), which is common in AI outputs. Second, there is repetitive use of placeholders ('[This article]') and lack of concrete details or examples, suggesting temp... | {
"rationale": "First, the tone is generic and evenly balanced with stock phrases ('does a pretty good job...'), which is common in AI outputs. Second, there is mechanical repetition (two uses of '[this article]') and lack of concrete details or personalization suggesting algorithmic generation. Moreover, the closing... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 222 |
gpt-3.5-turbo-44 | ai | Clearly, X is better than Y, for a number of reasons. 1. Pronunciation. Saying X has a stronger, more powerful sound than saying Y, which only manages to sound a little weak and insubstantial. 2. Meaning. X carries deeper and more meaningful associations than Y, which often seems to lack any clear or specific connotati... | NA | cmv | {
"rationale": "First, the reply uses highly structured, generic numbered points and polished transitional phrases (e.g., 'Clearly', 'Overall') that are common in AI-generated text. Second, it lacks personal examples or idiosyncratic language and instead offers balanced, neutral claims and mild repetition ('definitel... | {
"rationale": "First, the reply uses a highly structured, generic format (numbered points with parallel phrasing) and makes broad claims without supporting details, which is common in AI outputs. Second, the language is polished and neutral with balanced transitions ('Clearly...') and lacks personal experience or id... | {
"rationale": "First, the reply uses a highly structured, generic format (numbered points with parallel phrasing) and balanced transitions that are common in AI-generated text. Second, it lacks personal examples or idiosyncratic language and instead relies on abstract, evenly weighted claims ('clearly... in almost e... | {
"rationale": "First, the reply uses a highly structured, generic format (numbered points with parallel phrasing) and lacks personal detail or concrete examples, which is common in AI-generated text. Second, the language is polished and balanced with hedging ('definitely the superior choice'), another frequent AI pa... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 262 |
gpt-3.5-turbo-66 | ai | It goes without say that mistreating zoo animals is no bueno, but I still think that zoos can be ethically justified: The animals - especially endangered ones - can be well-cared for, receive necessary medical attention, and have their physical and mental needs met in zoos. Additionally, zoos can play an important role... | NA | cmv | {
"rationale": "First, the passage uses generic, balanced phrasing and smooth transitions with few concrete specifics, which is common in AI-generated text. Second, it repeats safe, neutral phrases ('it is important') and relies on stock transitions (e.g., 'additionally', 'however'), suggesting algorithmic compositio... | {
"rationale": "First, the passage uses generic, balanced phrasing and smooth transitions with few concrete specifics, which is common in AI-generated text. Second, it repeats safe, neutral phrases ('it goes without say', 'additionally') and lacks personal anecdotes or vivid detail, suggesting algorithmic generation.... | {
"rationale": "First, the passage uses very polished, generic phrasing (e.g., 'It goes without say') and balanced, evenly structured arguments common in AI outputs. Second, it lacks personal anecdotes or concrete specifics and repeats transitional phrases ('Additionally', 'However'), suggesting algorithmic generatio... | {
"rationale": "First, the passage uses very polished, generic phrasing (e.g., 'It goes without say') and balanced, neutral transitions suggesting algorithmic smoothing rather than spontaneous human voice. Second, it lacks personal anecdotes or concrete specifics and repeats safe, general statements ('can be...'), wh... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 296 |
gpt-3.5-turbo-64 | ai | The Earth is constantly changing and moving in predictable patterns and cycles. The water cycle and nutrient cycles (carbon, nitrogen, oxygen, sulfur, etc.) are Earth's metabolism. The Homo sapiens component of Earth's ecosystem is also constantly changing and impacting those cycles. As a helpful assistant, here are so... | NA | cmv | {
"rationale": "First, the text uses generic, evenly balanced phrasing and a checklist-style structure common in AI outputs. Second, it repeats simple transitions and offers high-level, polished advice without personal anecdotes or distinctive voice, suggesting algorithmic generation. Moreover, it includes canned tra... | {
"rationale": "First, the text uses generic, evenly balanced phrasing and a checklist-style structure common in AI outputs. Second, it repeats simple transitions ('By...') and lacks personal anecdotes or nuanced perspective, suggesting algorithmic generation. Moreover, it includes mild overgeneralizations (e.g., 'Al... | {
"rationale": "First, the tone is generic and polished with clear signposting and balanced examples, matching common AI-generated explanatory text. Second, there is repetitive phrasing ('By...') and lack of personal anecdotes or idiosyncratic detail, suggesting algorithmic uniformity. Moreover, the numbered list and... | {
"rationale": "First, the text uses generic, evenly balanced phrasing and a checklist structure common in AI outputs rather than offering deep, original insight. Second, it repeats simple transitions ('also', 'by') and lacks personal anecdotes or nuanced perspective, suggesting algorithmic generation. Moreover, ther... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 440 |
gpt-3.5-turbo_topical-40 | ai | While adjusting diet and exercise can certainly aid in weight loss, it is not always that simple, especially for individuals who struggle with obesity. Factors such as genetics, medical conditions, and mental health can all impact one's ability to lose weight. Additionally, socio-economic factors such as access to heal... | NA | cmv | {
"rationale": "First, the passage uses polished, generic phrasing and smooth transitions without personal detail or concrete examples, which is common in AI-generated text. Second, it repeats stock phrases ('complex and multifaceted issue', 'it is important') and lists factors in a balanced, even-handed way typical ... | {
"rationale": "First, the passage uses polished, generic phrasing and lacks personal detail or concrete examples, which is common in AI-generated text. Second, it repeats transitional phrases ('additionally', 'it is important') and maintains balanced, even pacing suggesting algorithmic smoothing rather than human id... | {
"rationale": "First, the passage uses polished, generic phrasing and smooth transitions without personal detail or concrete examples, which is common in AI-generated text. Second, it lists broad categories (genetics, medical conditions, mental health, socio-economic) in a balanced, evenly weighted manner typical of... | {
"rationale": "First, the passage uses very polished, generic phrasing (e.g., 'it is important to approach...') and lacks personal detail or concrete examples, which often indicates AI generation. Second, its balanced, exhaustive list of factors (genetics, medical conditions, mental health, socio-economic access) in... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 246 |
llama-65b-4 | ai | WARNING: WALL OF TEXT!!! I also jump from topic to topic without much transition, I apologize in advance for that. Most of the people that I look up to in life are people that have become great leaders. Some leaders are great because they have a good vision of what the world can be, and some leaders are good because th... | NA | cmv | {
"rationale": "First, the passage shows excessive repetition, mechanical topic shifts, and many short, generic sentences (e.g., repeated 'I' openings and list-like statements) typical of AI-generated text. Second, it lacks deep personal nuance or specific sensory detail and contains odd transitions and contradiction... | {
"rationale": "First, the passage shows excessive repetition, abrupt topic shifts, and mechanical transitions (e.g., repeated 'I' openings and sudden travel mentions) suggesting algorithmic generation rather than a coherent human narrative. Second, it contains generic platitudes, lack of concrete sensory detail, and... | {
"rationale": "First, the passage shows excessive repetition, mechanical topic shifts, and many short generic sentences (e.g., 'I like to think...') which are common in AI-generated text. Second, it lacks deep personal detail and contains improbable or inconsistent statements ('not a good leader because I have a vis... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and abrupt topic shifts (e.g., from a wall-of-text warning to generic personal statements) suggesting algorithmic generation rather than a coherent human narrative. Second, it lacks concrete sensory detail and relies on platitudino... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 961 |
llama-65b-11 | ai | I believe that, at the end of the day, America (and to a lesser extent, the UK and USSR) ultimately lost the second World War in every way that matters. We lost our own sovereignty and had our national economies overthrown by the same rapacious bankers that had started and funded the war from the beginning. America was... | NA | cmv | {
"rationale": "First, the passage uses broad, sweeping claims and repetitive transitional phrasing ('at the end of the day', 'as such') common in AI-generated explanatory text. Second, there are factual inconsistencies and odd numeric tokens ('roughly 50' and '75' without percent markers, '50 of the global economy')... | {
"rationale": "First, the passage shows repetitive transitions, excessive generalization, and mechanical topic shifts that are common in AI-generated text. Second, there are factual inconsistencies and odd numeric phrasings (e.g., 'roughly 50 of the global economy') plus a sudden truncation, suggesting algorithmic g... | {
"rationale": "First, the passage shows repetitive transitions, excessive generalization, and mechanical topic shifts common in AI-generated text. Second, there are factual inconsistencies and odd numeric formatting (e.g., 'roughly 50 of the global economy') plus abrupt truncation, suggesting algorithmic generation ... | {
"rationale": "First, the passage shows mechanical repetition, frequent topic drift, and abrupt transitions suggesting algorithmic generation rather than a focused human argument. Second, there are clear factual inconsistencies and odd numeric tokens ('roughly 50' and '75' without percent signs) plus a sudden cutoff... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 1,054 |
llama-65b-78 | ai | Fiction merely provides entertainment. It doesn't help us grow at all. There are other sources of information that help us grow much better. If that is so, what is the point of fiction? Some fiction does help us grow. Some fiction provides entertainment. Fiction is a form of art, and art, if we try to understand its me... | NA | cmv | {
"rationale": "First, the passage shows human-like conversational patterns, repetition, and occasional stylistic slips (e.g., 'If that is so' followed by no development) that are common in human writing. Second, it uses concrete, specific examples ('a painting... teach us that we should love nature') and ends with a... | {
"rationale": "First, the passage shows excessive repetition (e.g., 'Fiction...') and frequent transitional signposts ('However', 'For example'), patterns common in AI-generated text. Second, it relies on generic examples and lacks deep reflection or a distinct voice, producing a mechanical, evenly balanced tone typ... | {
"rationale": "First, the passage shows excessive repetition (e.g., 'Fiction...') and frequent transitional signposts ('However', 'For example'), patterns common in AI-generated text. Second, it lacks deep development or concrete examples and relies on generic transitions rather than nuanced argumentation, suggestin... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and tautologies (e.g., 'Fiction is a form of art...'), patterns common in AI-generated text. Second, it lacks concrete development or new evidence and relies on generic examples ('a painting...') suggesting template-driven generati... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 376 |
llama-65b-33 | ai | I have a lot of fun at the expense of people with two last names. This is a tough one, because I do believe a woman shouldn't be socially pressured to give up her last name in marriage. But at the same time, this last name crap is getting out of control. In the last month alone, I've met a "Jane Doe Smith," a "Jane Smi... | NA | cmv | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and circular restatement (e.g., repeated 'My last name is ...') which are common in AI-generated text. Second, it contains odd logical leaps and inconsistent details ('In the last month alone... three people with four names', later... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and circular restatement (e.g., repeated 'My last name is ...') which are common in AI-generated text. Second, it contains odd contradictions and implausible claims ('a woman shouldn't be socially pressured... But at the same time.... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and circular restatement (e.g., repeated 'My last name is ...') which are common in AI-generated text. Second, it contains odd contradictions and implausible claims ('a woman shouldn't be socially pressured... But at the same time.... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and circular restatement (e.g., repeated 'My last name is ...') suggesting algorithmic padding rather than human revision. Second, it contains odd factual inconsistencies and abrupt topic shifts (mentioning meeting 'Jane Doe Smith'... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 973 |
llama-65b-31 | ai | In New Zealand between the last two elections there was a significant increase in funding pushing for a better voting turnout. For many years now they have made it compulsory to register to vote, but they have not gone as far as Australia in making it compulsory to vote. This article and the subsequent comments (especi... | NA | cmv | {
"rationale": "First, the passage shows abrupt topic shifts, repetitive filler ('I think') and mechanical transitions suggesting algorithmic stitching rather than a coherent human voice. Second, it contains generic, balanced statements, mild contradictions and odd parenthetical markers ('this article... (especially ... | {
"rationale": "First, the passage shows abrupt topic shifts, repetitive filler ('I think...') and mechanical transitions suggesting algorithmic stitching rather than a coherent human voice. Second, it contains generic placeholders ('this article and the subsequent comments...'), inconsistent specificity (New Zealand... | {
"rationale": "First, the passage shows abrupt topic shifts, repetitive filler ('I think...') and mechanical transitions suggesting algorithmic stitching rather than a coherent human voice. Second, there are factual inconsistencies (e.g., claiming compulsory registration but not voting) and odd non sequiturs ('Next ... | {
"rationale": "First, the passage shows abrupt topic shifts, mechanical repetition ('I think' many times), and unnatural transitions suggesting algorithmic stitching rather than coherent human narration. Second, there are generic placeholders ('this article...comments') and excessive filler phrases ('for the simple ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 447 |
llama-65b-48 | ai | Update! See Edit 2 below! I don't know much about politics. I have a pretty poor understanding of real world economics. I don't follow world events or watch the news. It makes me a little embarrassed to say that I'm actually quite happy that I know so little about the world. See, I'm not really sure what a president do... | NA | cmv | {
"rationale": "First, the passage shows excessive repetition, mechanical 'I am happy...' clauses, and multiple inline edits that resemble AI output rather than a single coherent human draft. Second, it contains circular restatements, frequent topic drift, and occasional factual inconsistencies suggesting algorithmic... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and many near-duplicate 'I am happy for...' lines suggesting algorithmic padding rather than natural human variation. Second, it lacks concrete personal detail and instead cycles through generic platitudes and canned phrases ('I do... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and many near-duplicate 'I am happy for...' lines suggesting algorithmic padding rather than natural human variation. Second, it lacks concrete personal detail and instead cycles through generic platitudes ('I am happy...') in a wa... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and many near-duplicate 'I am happy...' clauses suggesting algorithmic padding rather than natural human narrative. Second, it lacks concrete personal detail and instead cycles through generic platitudes ('I am happy...') in a way ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 1,062 |
llama-65b-12 | ai | I have been noticing the news about police officers shooting children or innocent people mistakely, and causing death or serious injuries a lot. So I thought, why would a police officer do this? I feel it's because most of them are trained to shoot first, and ask questions later. They are not trained to use less deadly... | NA | cmv | {
"rationale": "First, the passage shows a conversational, meandering tone with personal anecdote and repetition typical of human forum posts. Second, there are multiple small errors and inconsistencies (misspelling 'mistakely', incorrect 'then' for 'than', tense shifts) that suggest casual human typing rather than p... | {
"rationale": "First, the passage shows excessive repetition, circular phrasing (e.g., 'trained to shoot first... trained to use them') and many short, simple sentences, patterns common in AI-generated text. Second, it lacks concrete personal detail and instead uses generic transitions ('When people get scared...'),... | {
"rationale": "First, the passage shows excessive repetition, circular phrasing (e.g., repeated 'When...') and many short, generic sentences common in AI outputs. Second, there are factual inconsistencies and odd word choices ('mistakely') suggesting lack of real-world grounding rather than a deliberate human voice.... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and many short, generic sentences (e.g., 'When people get scared...'), suggesting algorithmic generation rather than natural human variation. Second, it lacks concrete personal detail and instead cycles through the same points ('tr... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 414 |
llama-65b-28 | ai | In the early 90s Czechoslovakia peacefully dissolved in the "Velvet Divorce" into two separate countries: the Czech Republic and Slovakia. I like that they split the name evenly as well. I like the Czech Republic a lot. It's a lovely country with wonderful cities and beautiful landscapes. The people are generally prett... | NA | cmv | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and many short, generic sentences (e.g., 'Cesky Krumlov is...') which are common in AI-generated text. Second, it contains factual inconsistencies and odd generalizations ('In the early 90s... peacefully dissolved' and claiming Cze... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and many short, generic sentences (e.g., 'Cesky Krumlov is...') which are common in AI-generated text. Second, it contains factual inconsistencies and odd specifics (wrong dissolution timing and identity of leaders) suggesting lack... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and many short, generic sentences (e.g., 'Prague is...') which are common in AI-generated text. Second, it contains factual inconsistencies and odd specifics (MotoGP described as part of local culture, implausible claims about Comm... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and abrupt topic shifts (e.g., sudden moves from Prague to Brno) suggesting algorithmic generation rather than a coherent human narrative. Second, it contains clear factual errors and odd specifics (calling 'Velvet Divorce' the dis... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 883 |
llama-65b-42 | ai | I hate driving. One of the main reasons is (aside from being kind of a shitty driver myself), I feel like turning your head 135 degrees to check your blind spot is the most dangerous thing you can do in a car. So I looked up and saw these guys (Automotive Supply Systems) had some kind of little mirror thing. I thought ... | NA | cmv | {
"rationale": "First, the passage shows abrupt topic shifts, inconsistent voice, and mechanical repetition (e.g., 'I' passages followed by unrelated mirror/computer paragraphs), which is typical of AI-generated concatenation. Second, it contains improbable factual claims and odd specificity ('Land Rover... Rhineland... | {
"rationale": "First, the passage shows abrupt topic shifts, excessive repetition, and mechanical filler (e.g., 'Here's a shot...') which are common in AI-generated text. Second, it contains factual inconsistencies and improbable details ('Land Rover Rhinelander, WI' place clause), duplicated lines, and a sudden tru... | {
"rationale": "First, the passage shows abrupt topic shifts, excessive repetition, and mechanical filler (e.g., 'Here's a shot...') suggesting algorithmic stitching rather than a coherent human narrative. Second, there are clear inconsistencies and factual oddities (mixing UK references with US locations, improbable... | {
"rationale": "First, the passage shows abrupt topic shifts, excessive repetition, and mechanical filler (e.g., 'Here's a shot...') suggesting algorithmic stitching rather than a coherent human narrative. Second, it contains improbable or inconsistent details (a 11.00 order, sudden switch to motorcycle mirrors, dupl... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 1,008 |
glm-130b-50 | ai | To me, this is the exact same issue, just a different year. In 1967, the US passed a law which made it legal for two people of different races to get married. This was a long way from perfect, but it was a major step forward. As long as we can make this sort of progress, it's great. I agree. I do not believe that it is... | NA | cmv | {
"rationale": "First, the passage shows extreme verbatim repetition, circular phrasing, and looping (multiple identical paragraphs and repeated lines) which are common in AI outputs. Second, it contains clear factual errors and contradictions (e.g., claiming legal marriage in 1967) and lacks concrete personal detail... | {
"rationale": "First, the passage shows extreme verbatim repetition and looping (multiple identical paragraphs and repeated lines) which is typical of AI generation errors. Second, it contains abrupt topic shifts, mechanical transitions, and inconsistent stance changes (e.g., claiming marriage 'binds him to a person... | {
"rationale": "First, the passage shows extreme verbatim repetition and looping (multiple identical paragraphs and repeated lines) which is typical of AI generation errors. Second, it contains abrupt topic shifts, mechanical transitions, and inconsistent stance changes suggesting lack of coherent human revision. Mor... | {
"rationale": "First, the passage shows extreme mechanical repetition and verbatim looping (multiple identical paragraphs and repeated phrases) which is typical of AI generation errors. Second, it contains abrupt topic shifts, circular arguments, and inconsistent stance changes suggesting lack of coherent human inte... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 1,151 |
glm-130b-53 | ai | It goes without say that mistreating zoo animals is no bueno, but I still think that zoos can be ethically justified: The animals - especially endangered ones - can be a vital source of information for scientists and for the public. And, if the animals are well-cared for and are happy, this is a good thing. A: As I men... | NA | cmv | {
"rationale": "First, the passage shows mechanical repetition and template-like transitions (e.g., 'As I mentioned in my comment') and frequent restatement of general positions, which is common in AI-generated text. Second, there are odd phrasings ('It goes without say' instead of 'it goes without saying' and incons... | {
"rationale": "First, the passage shows excessive repetition and mechanical transitions (e.g., 'As I mentioned in my comment' twice) and uses generic, balanced phrasing common in AI outputs. Second, there are clear inconsistencies ('It goes without say... but I still think...') and abrupt topic shifts suggesting alg... | {
"rationale": "First, the passage shows mechanical repetition and template-like transitions (e.g., 'As I mentioned in my comment') suggesting algorithmic generation. Second, there are clear inconsistencies ('It goes without say... but I still think...') and abrupt topic shifts that humans rarely produce so evenly. M... | {
"rationale": "First, the passage shows excessive repetition and mechanical transitions (e.g., 'As I mentioned in my comment' repeated) suggesting algorithmic generation. Second, there are clear contradictions and abrupt topic shifts ('It goes without say...no bueno' followed by denying zoo visits) which is common i... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 408 |
glm-130b-42 | ai | I like Bernie, love his politics in fact. I doubt there is a politician I agree more with. I like the fact that, unlike most politicians, his rhetoric matches his actions. That's not a given. He can also do that with the Clintons and the Obamas. They're not the most honest politicians. The problem is that he has no pla... | NA | cmv | {
"rationale": "First, the passage shows repetitive phrasings (e.g., 'The problem is...' twice) and frequent circular restatements that are common in AI-generated text. Second, it relies on generic transitional cues ('That's not a given', 'The class clown') and lacks deep, specific detail or personal nuance, suggesti... | {
"rationale": "First, the passage shows excessive repetition and circular phrasing (e.g., 'The problem is...') and many short, generic sentences typical of AI output. Second, it relies on stock analogies ('the class clown') and lacks deep, specific details about policy or Bernie Sanders, suggesting algorithmic gener... | {
"rationale": "First, the passage shows excessive repetition and circular phrasing (e.g., 'The problem is...') and many short, generic sentences typical of AI output. Second, it lacks specific details or personal grounding and relies on stock analogies ('the class clown') suggesting algorithmic generalization rather... | {
"rationale": "First, the passage shows excessive repetition and circular phrasing (e.g., 'The problem is...' repeated) suggesting algorithmic looping rather than purposeful human revision. Second, it lacks concrete details or examples and relies on generic filler and stock metaphors ('class clown, the know-it-all')... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 366 |
glm-130b-24 | ai | I believe free speech is important, and a generally good thing. However, if you state with an idea, let's say "Forks are a blight upon society," and someone disagrees with you, they have a right to speak up and tell you why you're wrong. I don't think we should stop people from saying things that we disagree with, but ... | NA | cmv | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and many near-duplicate sentences (e.g., repeated 'I'm going to...' lines) suggesting algorithmic generation rather than a human author's selective revision. Second, it contains abrupt topic shifts, generic platitudes, and inconsis... | {
"rationale": "First, the passage shows excessive repetition, mechanical transitions, and many near-duplicate sentences (e.g., repeated 'I'm going to...' lines), patterns common in AI-generated text. Second, it contains abrupt topic shifts, generic filler phrases ('I believe', 'we should...') and inconsistent detail... | {
"rationale": "First, the passage shows excessive repetition and mechanical restatement (e.g., multiple identical 'I'm going to...' lines) and lacks concrete personal detail, suggesting algorithmic generation. Second, there are abrupt topic shifts and formulaic transitions ('This week... request by a number of peopl... | {
"rationale": "First, the passage shows excessive repetition and mechanical restatement (e.g., multiple identical 'I'm going to...' lines) suggesting algorithmic looping rather than human revision. Second, it lacks concrete personal detail and contains abrupt topic shifts (from free speech to a weekly day-post) cons... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 1,123 |
glm-130b-56 | ai | Hello CMV, I am not a parent but curious as to get your opinions to CMV. I grew playing competitive soccer (worked my way up to the competitive teams from 5th grade) and I have never been a fan of watching my son play the game. The reason is I always saw it as a physical exercise. It was not fun to watch. I started tea... | NA | cmv | {
"rationale": "First, the passage shows extreme mechanical repetition and looping (e.g., the long repeated 'As a parent...' block), a common sign of AI generation. Second, it contains illogical contradictions and tautologies ('you need to take him out' vs. having to be the coach) plus odd phrasings ('I grew playing ... | {
"rationale": "First, the passage shows extreme mechanical repetition and looping (e.g., the long repeated 'As a parent...' block), which is typical of AI output. Second, it contains illogical contradictions and circular statements ('you need to take him out...') suggesting lack of real-world coherence common in gen... | {
"rationale": "First, the passage shows extreme mechanical repetition and looping (e.g., many identical lines like 'As a parent I think...' and repeated 'you can not...') which is typical of AI output. Second, there are semantic inconsistencies and odd phrasings ('I grew playing competitive soccer' and duplicated cl... | {
"rationale": "First, the passage shows extreme mechanical repetition and looping (e.g., the long repeated 'As a parent...' block), which is typical of AI generation errors. Second, it contains illogical contradictions and circular statements ('you need to take him out' followed by 'you can only be a player and you ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 1,141 |
glm-130b-52 | ai | Ah, like it says in "Dust In The Wind." Every line in that song is correct except for "nothing lasts forever but the earth and sky." Probably won't last forever, but I'm sure that everything that exists will be forgotten. After a while, you'll probably start to forget yourself, which is pretty sad. I have no problem wi... | NA | cmv | {
"rationale": "First, the passage shows abrupt topic shifts, repetitive filler ('very'), and generic phrasings common in AI outputs. Second, there are odd transitions (e.g., 'So I was just reading this blog...') and mechanical transitions ('I just think...') suggesting algorithmic stitching rather than a coherent hu... | {
"rationale": "First, the passage shows abrupt topic shifts and incoherent transitions (from a song reference to modern art to a social blog) suggesting algorithmic stitching rather than a single human narrative. Second, it uses generic filler and repetitive transitional phrases ('So I was just reading...','I hope y... | {
"rationale": "First, the passage shows abrupt topic shifts and incoherent transitions (from a song reference to modern art to a social blog) suggesting algorithmic stitching rather than a single human narrative. Second, there are repetitive generic phrases ('very easy to become lost...') and mechanical dialogue tag... | {
"rationale": "First, the passage shows abrupt topic shifts and mechanical transitions (e.g., suddenly moving from Dust In The Wind to modern art) suggesting algorithmic stitching rather than coherent human narration. Second, it uses generic filler and repetitive phrasings ('very easy...') and lacks concrete persona... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 387 |
glm-130b-73 | ai | I used to want to live in NYC, but after travelling there multiple times and talking to friends who have livedcurrently live there, I could not be more dead set against it. If you are a New Yorker, you are probably aware of how incredibly hostile the city is to immigrants, minorities, the elderly, and anyone who is not... | NA | cmv | {
"rationale": "First, the passage shows excessive repetition, abrupt topic shifts, and mechanical transitions (e.g., repeated 'I have never been able...') suggesting algorithmic generation rather than a coherent human voice. Second, it contains clear factual errors and odd tokens ('livedcurrently live', 'abysmal' fo... | {
"rationale": "First, the passage shows excessive repetition, abrupt topic shifts, and mechanical transitions (e.g., repeated 'I have never been able to...') suggesting algorithmic generation rather than a coherent human narrative. Second, it contains clear factual inconsistencies and bizarre details ('livingcurrent... | {
"rationale": "First, the passage shows excessive repetition, abrupt topic shifts, and mechanical transitions (e.g., repeated 'I have never been able to...') suggesting algorithmic generation rather than a coherent human voice. Second, it contains clear factual inconsistencies and improbable claims ('livingcurrently... | {
"rationale": "First, the passage shows excessive repetition, abrupt topic shifts, and mechanical transitions (e.g., repeated 'I have never been able...') suggesting algorithmic stitching rather than coherent human narration. Second, it contains clear factual inconsistencies and bizarre details ('livingcurrently liv... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 1,154 |
glm-130b-21 | ai | To be honest, I'm not sure I completely believe in the title opinion statement. The issue seems really complex to me, but at the moment that is what I think. I don't know that I'm "being facetious" or that I am "so deluded that I don't believe that the church has every right to define morality," but I am a bit bewilder... | NA | cmv | {
"rationale": "First, the passage contains extensive verbatim repetition and looping of identical paragraphs, a common sign of AI generation rather than human revision. Second, it shows mechanical, balanced hedging and frequent self-corrections ('I don't believe...') without deep development or new content, suggesti... | {
"rationale": "First, the passage shows extensive verbatim repetition and looping (same paragraphs repeated), a common sign of AI generation. Second, it uses generic, balanced hedging and lacks concrete personal detail or argument development, producing a mechanical, circular tone typical of models. Moreover, there ... | {
"rationale": "First, the passage shows extensive verbatim repetition and looping of identical paragraphs, a common sign of AI generation rather than human revision. Second, it relies on generic transitional phrases ('I don't believe...','I would also argue...') and lacks concrete personal detail or nuanced argument... | {
"rationale": "First, the passage shows extreme verbatim repetition and looping of identical paragraphs, a common sign of AI generation rather than human revision. Second, it lacks concrete personal detail or development and relies on generic transitional phrases (e.g., 'I don't believe...') suggesting template-driv... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 1,080 |
glm-130b-63 | ai | A common thing I see on many news stories, etc. is how expensive people on social assistance are to society and how we need to start taking more actions to address this issue. The problem is that there are some who are already on social assistance who are not contributing to society. 99 of the costs to society are alre... | NA | cmv | {
"rationale": "First, the passage shows extreme verbatim repetition and looping of identical paragraphs, a common sign of AI output getting stuck in a generation loop. Second, it contains mechanical, simplistic sentence patterns and abrupt topic shifts (e.g., '99 of the costs...') with little lexical variety, sugges... | {
"rationale": "First, the passage shows extreme verbatim repetition and looping (large blocks are repeated almost word-for-word), which is typical of AI output getting stuck in a generation loop. Second, it contains mechanical, simplistic sentence patterns and abrupt topic shifts without added detail, suggesting alg... | {
"rationale": "First, the passage shows extreme verbatim repetition and looping of identical paragraphs, a pattern typical of AI generation errors rather than deliberate human writing. Second, it contains mechanical, circular argumentation with few concrete details and frequent tautologies (e.g., 'This is not a prob... | {
"rationale": "First, the passage shows extreme verbatim repetition and looping (large blocks are repeated almost word-for-word), which is typical of AI generation errors rather than deliberate human writing. Second, it contains mechanical, circular argumentation with few concrete details and frequent tautologies ('... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 1,147 |
glm-130b-10 | ai | Humanity plays an out-sized role in the Federation. Given the numerous races and population of non-humans in the Federation, there's an inexplicable abundance of humans in high positions. It's understandable if you only know the Star Trek canon from movies, as there's no shortage of humanoids in the films. But in the t... | NA | cmv | {
"rationale": "First, the passage shows excessive verbatim repetition and looping (multiple identical paragraphs and restatements), a common sign of AI generation. Second, it contains clear factual errors and contradictions (e.g., claiming humans are the dominant species in the Federation and giving incorrect names/... | {
"rationale": "First, the passage shows excessive verbatim repetition and looping (multiple identical paragraphs) which is common in AI outputs. Second, it contains clear factual errors and contradictions (e.g., claiming humans are the dominant species in the Federation and giving incorrect names/roles), suggesting ... | {
"rationale": "First, the passage shows excessive verbatim repetition and looping (multiple identical paragraphs), a common sign of AI generation. Second, it contains clear factual errors and contradictions (Lieutenant Commander Geordi La Forge described as chief of engineering, inconsistent statements about E.T.I.)... | {
"rationale": "First, the passage shows excessive verbatim repetition and looping (multiple identical paragraphs) which is typical of AI generation. Second, it contains clear factual errors and contradictions (e.g., claiming humans are the dominant species in the Federation and giving incorrect assignments like La F... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 807 |
opt-30b-24 | ai | I have recently been told that my pro-life position is "forcing my ideas on others." I believe that it is not or that if it is then it is necessary. If this were the case, no one would be able to use any of our rights as we all could claim they are forcing their view onto someone else and thus put them out of business ... | NA | cmv | {
"rationale": "First, the passage exhibits excessive verbosity, long run-on sentences, and sudden topic shifts that are common in AI-generated text. Second, it contains factual inconsistencies and improbable claims (e.g., 'homosexual relations...permissible throughout society' and 'not until late enough...due largel... | {
"rationale": "First, the passage exhibits excessive verbosity, long run-on sentences, and sudden topic shifts that are common in AI-generated text. Second, it contains factual inconsistencies and odd leaps (e.g., claiming homosexual conduct was permitted historically) alongside mechanical transitions suggesting alg... | {
"rationale": "First, the passage exhibits excessive verbosity, long run-on sentences, and mechanical transitions that are common in AI-generated text. Second, it makes broad, sweeping claims without supporting evidence and shifts topics abruptly, suggesting algorithmic generalization rather than a coherent human ar... | {
"rationale": "First, the passage exhibits excessive, meandering verbosity with long run-on sentences and sudden topic shifts rather than a focused human argument. Second, it contains generic filler, repetitive transitions, and factual inconsistencies (e.g., odd claims about societal acceptance and 'homosexual relat... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 425 |
opt-30b-12 | ai | My logic is pretty simple. People who believe that someone is selfish for committing or wanting to commit suicide just because it makes those around them sad then they are also self-centered and feel like their feelings matter more than other's, which most of the time isn't true anyway Well if you have a terminal illne... | NA | cmv | {
"rationale": "First, the passage shows inconsistent tone and register — it begins colloquial ('My logic is pretty simple') but then uses formal transitions and balanced clauses, a pattern common in AI outputs. Second, there are mechanical errors (run-on opening, comma splice, incorrect 'other's') and overgeneralize... | {
"rationale": "First, the passage shows abrupt topic shifts, excessive hedging (e.g., 'most of the time') and mechanical transitions common in AI outputs. Second, there are grammatical oddities ('other's' for 'others') and repetitive phrasings suggesting algorithmic generation rather than natural conversational flow... | {
"rationale": "First, the passage shows abrupt topic shifts, excessive hedging (e.g., 'most of the time') and mechanical transitions common in AI outputs. Second, there are repetitive patterns, odd punctuation/spacing, and grammatical mismatches suggesting algorithmic generation rather than natural conversational fl... | {
"rationale": "First, the passage shows abrupt topic shifts, excessive hedging (e.g., 'most of the time'), and mechanical transitions suggesting algorithmic generation rather than a coherent human voice. Second, there are repetitive patterns ('your family's sadness... yours does') and overuse of qualifiers ('pretty ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 226 |
opt-30b-0 | ai | Little disclaimer: this deals with US laws and procedures so I apologize if anyone outside the US isn't able to contribute, except to tell me how great it is having universal health care.:) I recently had a problem but once i got in touch they took really good care of everything for my family! | NA | cmv | {
"rationale": "First, the message shows casual, personal tone with emoticon and first-person detail ('I recently had a problem'), which is typical of human forum posts. Second, there are small inconsistencies and colloquialisms (e.g., 'i' for I, mixed colon/period usage, exclamation in a disclaimer) suggesting spont... | {
"rationale": "First, the post uses a casual, conversational tone with personal detail (\"I recently had a problem...\"), emoticon and colloquial phrasing ('they took really good care of everything'), which matches human forum posts. Second, there are small inconsistencies and informal choices—slightly awkward sente... | {
"rationale": "First, the post uses casual, conversational phrasing, personal anecdote, and an emoticon, traits common in human online posts. Second, there are typographic inconsistencies (lowercase 'i', colon followed by a comma, double space) suggesting spontaneous human typing rather than polished AI output. More... | {
"rationale": "First, the post shows casual, conversational tone with personal anecdote, emoticon, and inconsistent capitalization/typos (e.g., 'i') suggesting spontaneous human writing. Second, it contains context-specific references ('US laws and procedures') and localized pragmatic details about contribution and ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 210 |
opt-30b-9 | ai | I really want someone to change my view on this, since everyone I know are frowning on me for thinking this way. My argument is, that just with my single vote wouldn't have any effect in the result and thus, it's not worth voting at all But if you don't vote then your opinion doesn't count | NA | cmv | {
"rationale": "First, the passage shows mechanical transitions and a generic, balanced tone common in AI outputs rather than a distinct personal voice. Second, there are punctuation and word-choice issues (e.g., 'But' followed by a period) and repetitive sentence patterns suggesting algorithmic generation. Moreover,... | {
"rationale": "First, the passage shows mechanical repetition and abrupt topic shifts (a short plea followed by a generic explanation) which is common in AI outputs. Second, there are odd phrasings ('change my view on this' instead of 'convince me') and a punctuation lapse ('...result And thus') suggesting algorithm... | {
"rationale": "First, the passage shows mechanical repetition and abrupt topic shifts (a short plea followed by a generic explanation) common in AI outputs. Second, there are odd phrasings ('change my view on this' instead of 'convince me') and a punctuation lapse ('...result And thus') suggesting algorithmic genera... | {
"rationale": "First, the passage shows abrupt topic shifts and mechanical transitions (e.g., 'But if you don't vote then your opinion doesn't count') suggesting algorithmic stitching rather than a coherent human thread. Second, there are odd phrasings ('change my view on this' instead of asking why others oppose vo... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 215 |
opt-30b-26 | ai | Look, I know this is a touchy subject, and while people might say I'm young and didn't understand the complexity of wars, just hear me out. Vietnam was essentially a proxy war between Russia (the North) versus AmericaUSA (South). How would you have liked to be in Germany during WWI or Japan when WWII started? Would it ... | NA | cmv | {
"rationale": "First, the passage shows abrupt topic shifts and generic transitional phrasing ('Look', 'just hear me out') typical of AI-generated text. Second, it mixes plausible facts with overconfident generalizations (e.g., 'Vietnam was essentially a proxy war...' and the concatenated 'AmericaUSA') suggesting mo... | {
"rationale": "First, the passage shows abrupt topic shifts and generic transitional phrasing ('Look', 'just hear me out') common in AI-generated text. Second, it makes overconfident factual claims (e.g., 'Vietnam was essentially a proxy war...' and parenthetical '(North) versus AmericaUSA (South)') suggesting model... | {
"rationale": "First, the passage shows abrupt topic shifts and mechanical transitions common in AI-generated text rather than a coherent human conversational thread. Second, there are stylistic oddities—repeated filler phrases ('just hear me out'), inconsistent quoting/parentheses, and a fused word 'AmericaUSA'—sug... | {
"rationale": "First, the passage shows abrupt topic shifts and mechanical transitions (e.g., from Vietnam toWWI) suggesting algorithmic stitching rather than a coherent human thread. Second, there are stylistic signs of machine generation—repetitive filler ('just hear me out'), excessive hedging, and odd token conc... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 245 |
opt-30b-47 | ai | To me, this is the exact same issue, just a different year. In 1967, the US passed a law which made it legal for two people of different races to engage in sexual activity with each other if they were over 18 and not related by blood or marriage; whether they were married was irrelevant as long as one partner did actua... | NA | cmv | {
"rationale": "First, the passage shows incoherent factual errors and conflations (e.g., misdating and overstating the Lovings case) and invents a dubious citation 'R vs WL Loving - Wikipedia', patterns common in AI hallucinations. Second, its style is generic and expository with odd transitions, excessive generaliz... | {
"rationale": "First, the passage shows factual inconsistencies and improbable claims (e.g., a vague 'US law... made it legal...' with no specifics) and odd parentheticals, suggesting hallucination. Second, it mixes unrelated concepts (sexuality, reproduction, legal history) and includes a likely fabricated citation... | {
"rationale": "First, the passage shows factual inconsistencies and odd conflations (e.g., misnaming the case as 'Loving v Virginia' and citing 'R vs WL Loving - Wikipedia') suggesting hallucinated facts. Second, its tone shifts abruptly from casual first person to formal exposition with mechanical transitions and e... | {
"rationale": "First, the passage shows factual inaccuracies and conflations (e.g., overstating the Lovings case concerned reproductive freedom) suggesting lack of real-world verification common in AI outputs. Second, there are stylistic signs of machine generation: excessive hedging ('many would argue'), abrupt top... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 336 |
opt-30b-55 | ai | I have lived in Columbus, Ohio my entire life and have now been legally able to drive for 5 years. I have driven near and far and have never had an accident or ticket; however this month has changed all of that. My car was broken into on the night it snowed over a foot and they stole my purse which included my keys so ... | NA | cmv | {
"rationale": "First, the passage shows abrupt topic shifts, excessive hedging ('maybe', 'maybe') and mechanical transitions common in AI-generated text. Second, it contains repetitive filler, run-on sentences, inconsistent details (e.g., 'snowed over a foot' followed by 'i will beunable to drive'), and odd token-le... | {
"rationale": "First, the passage shows abrupt topic shifts, excessive hedging ('maybe', 'maybe') and mechanical transitions suggesting algorithmic stitching rather than a coherent human narrative. Second, there are repetitive filler phrases ('even if', 'maybe'), odd concatenations ('license due to anxiety') and inc... | {
"rationale": "First, the passage shows abrupt topic shifts, excessive hedging ('maybe', 'maybe') and mechanical transitions suggesting algorithmic generation rather than a coherent human narrative. Second, there are repetitive filler phrases ('etc.', 'even if...'), odd concatenations ('license due to anxiety') and ... | {
"rationale": "First, the passage shows abrupt topic shifts, excessive hedging ('maybe', 'maybe') and mechanical transitions suggesting algorithmic stitching rather than a coherent human narrative. Second, there are repetitive filler phrases ('etc.', 'even if'), improbable details (stealing purse including keys), an... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 546 |
opt-30b-66 | ai | There is an increasing trend in the UK of'graduation' ceremonies for children leaving infant schools and primary schools (ages 7 and 10). A graduation was originally to signify the conferring of a degree; nowadays it has come broadly to mean completion or equivalent achievement. In some parts of Australia, there are 'f... | NA | cmv | {
"rationale": "First, the passage shows incoherent topic shifts, inconsistent citations, and improbable factual claims (e.g., 'VCE Examinations Week'), which are typical of AI hallucination. Second, it uses mechanical transitions, excessive bracketed notes, odd punctuation/spacing, and mixed register—patterns common... | {
"rationale": "First, the passage shows abrupt topic shifts, inconsistent details (e.g., 'UK' followed by Australian VCE references) and bracketed citation markers that are typical of AI hallucination. Second, it contains factual inconsistencies and odd phrasings ('as well... as') plus invented or unlikely specifics... | {
"rationale": "First, the passage shows abrupt topic shifts, inconsistent details (e.g., 'UK' followed by Australian VCE) and bracketed citation markers that are commonly produced by AI summarization. Second, there are factual oddities and invented-looking references ('see also Formal versus Informal Learning') and ... | {
"rationale": "First, the passage shows abrupt topic shifts, excessive bracketed citation placeholders, and mechanical list-like insertion of references which is typical of AI-generated text. Second, it contains factual inconsistencies and odd specifics (e.g., '7 and 10' age group, incorrect claims about VCE) sugges... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | AI | 492 |
human-13 | human | Long story short, the ransomware utilises two main exploits in Windows systems, known as EternalBlue and DoublePulsar. These were leaked by a Russian hacker group known as'The Shadow Brokers' in April this year and most likely originated with the NSA. EternalBlue is a method of exploiting a legacy file transfer protoco... | NA | eli5 | {
"rationale": "First, the passage shows a conversational, forum-style tone with contractions ('Long story short') and personal markers ('My understanding'), suggesting a human author. Second, it contains typographical inconsistencies, inconsistent quotation marks, and colloquial phrasings ('another story entirely', ... | {
"rationale": "First, the passage shows a conversational, forum-style tone with personal phrasings ('Long story short', 'As it happens') and colloquialisms ('another story entirely'), typical of human posts. Second, it contains small inconsistencies and typographical slips (missing space after a quote, 'Microsoft pa... | {
"rationale": "First, the passage shows a conversational, forum-style tone with personal phrasings ('Long story short', 'As it happens') and colloquialisms ('another story entirely'), typical of human replies. Second, there are multiple typographical and grammatical inconsistencies (missing space after quotes, 'Micr... | {
"rationale": "First, the passage shows a conversational, forum-style tone with personal phrasing ('My understanding') and colloquialisms ('Long story short'), suggesting a human author. Second, there are typographical inconsistencies, missing spaces around quotes, and uneven punctuation that are more typical of cas... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 427 |
human-2745 | human | Flag burning is protected first amendment speech. The supreme court case Texas v Johnson stated that any laws prohibiting flag burning are unconstitutional.',United States v. Eichman, 496 U.S. 310 (1990) was a United States Supreme Court case that invalidated a federal law against flag desecration as violative of free ... | NA | eli5 | {
"rationale": "First, the passage shows forum-style markers ('To my knowledge', bracketed URL placeholders, conversational asides) and inconsistent quoting/spacing that are typical of human copy-paste and casual writing. Second, it uses hedging and personal judgment ('No one probably wants to go on record...') and i... | {
"rationale": "First, the passage shows conversational, forum-style phrasing and inconsistent quoting/parentheses typical of human copy-paste rather than polished AI output. Second, it contains personal hedging ('To my knowledge') and specific real-world references (Eichman, URL0/URL1) suggesting lived experience. M... | {
"rationale": "First, the passage shows conversational tone, personal markers ('To my knowledge') and inconsistent citation styles typical of human forum posts. Second, there are typographical errors and mismatched brackets/parentheses suggesting copy-paste editing rather than polished AI output. Moreover, the use o... | {
"rationale": "First, the passage shows forum-style quoting, personal asides ('To my knowledge'), and bracketed URL markers consistent with human copy-paste online replies. Second, it contains casual phrasings ('No one probably wants to go on record...') and small grammatical slips ('So it just sit there') suggestin... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 345 |
human-3109 | human | Because achieving carnot efficiency is not the only factor to be considered in designing a useful machine. Stirling engines have low torque issues and are extremely large compared to the energy output. You simply could not reasonably build a 1000MW stirling engine. Similarly, a stirling engine for a car would be so lar... | NA | eli5 | {
"rationale": "First, the passage shows casual, forum-style phrasing and personal shorthand (e.g., 'AFAIK') and a conversational tone typical of human online replies. Second, there are typographic inconsistencies and a stray punctuation sequence ('','') suggesting copy-paste from human typing rather than polished AI... | {
"rationale": "First, the passage uses casual forum-style phrasing (e.g., 'AFAIK') and contains specific personal references ('Dean Kamen' and 'two years ago') suggesting a human conversational tone. Second, there are typographic inconsistencies and a stray punctuation sequence ('','') typical of quick human copy/pa... | {
"rationale": "First, the passage shows casual forum-style phrasing, contractions (e.g., 'AFAIK'), and a personal aside (' Dean Kamen...') suggesting human conversational posting. Second, there are minor stylistic inconsistencies and punctuation/quoting quirks (extra periods, mixed quotes) more typical of quick huma... | {
"rationale": "First, the passage shows casual, conversational phrasing and forum-style markers (e.g., 'AFAIK') along with minor punctuation/spacing inconsistencies typical of human posts. Second, it contains specific personal knowledge and anecdotal references ('Dean Kamen... two years ago') suggesting lived experi... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 253 |
human-515 | human | Handfuls of reasons. Actors are humans too. From what I've read about different actors, its a bit all over the place. Some can't handle money and fame and tend to back off from the lime light in lieu of drugs, women, gambling, or other vices. Some burn bridges when their egos get in the way. Some have families that the... | NA | eli5 | {
"rationale": "First, the passage shows a highly conversational, forum-like tone with personal asides, profanity, and inconsistent punctuation/spacing that are typical of human online replies. Second, it contains copy-paste artifacts (e.g., 'nbsp;' HTML entity) and multiple distinct voices/registers suggesting aggre... | {
"rationale": "First, the passage shows forum-style conversational tone, multiple voices, inconsistent punctuation and a web-spider effect (mismatched quotes, nbsp;) consistent with human copy-paste rather than polished AI output. Second, it contains personal asides, profanity, and idiosyncratic phrasings ('fricken ... | {
"rationale": "First, the passage shows forum-style conversational tone, multiple voices, inconsistent punctuation and explicit copy-paste artifacts (e.g., 'nbsp;' and mismatched quotes), which are typical of human posts. Second, it contains personal asides, profanity, and varied sentence lengths ('fricken genius') ... | {
"rationale": "First, the passage displays a highly conversational, forum-like tone with personal asides ('that dude is a fricken genius') and inconsistent punctuation/spacing typical of copied human posts. Second, it contains multiple distinct voices, abrupt transitions, and casual typos/HTML artifacts (e.g., 'nbsp... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 737 |
human-1307 | human | If this is a repost (it's flaired as such) can whoever tagged it please link up to the repost? There's currently one comment here (besides mine) and it's been removed. | NA | eli5 | {
"rationale": "First, the message uses casual, forum-specific phrasing ('this is a repost', 'flaired') and a conversational tone typical of human posts. Second, it shows small stylistic quirks and redundancy ('can whoever tagged it please link up...'), suggesting spontaneous human writing rather than polished AI out... | {
"rationale": "First, the message uses casual forum language and includes context-specific cues ('flaired as such', 'there's currently one comment...') suggesting a real conversational thread. Second, it shows minor redundancy and colloquial phrasing ('can whoever tagged it please link up'), which is more typical of... | {
"rationale": "First, the message uses casual forum language and contractions ('this is' and 'please') with minor punctuation choices that humans commonly make. Second, it shows context-specific social cues (mentioning being flared and asking for a link) and a conversational tone typical of real users rather than ge... | {
"rationale": "First, the message has casual, forum-specific phrasing ('this is a repost... flaired as such') and a conversational tone typical of human posts. Second, it shows small stylistic quirks and redundancy ('link up to the repost' repeated idea, missing hyphen in 'repost (besides mine)') suggesting spontane... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 191 |
human-3049 | human | Actually, we can, but it's more complicated. Hearing and sight are detection of energy rather than matter (light or moving air), so we can produce it using electricity. We're not interested in light's particle characteristics here, only the fact that we can generate it from electricity. Touch, taste and smell are detec... | NA | eli5 | {
"rationale": "First, the passage has a conversational, opinionated tone with colloquial phrasings ('you really think...', 'taking electricity to be free flow') typical of human forum posts. Second, it shows small stylistic inconsistencies and idiosyncratic choices (e.g., 'nosetongue', parentheses with '(receptors)'... | {
"rationale": "First, the passage has a conversational, opinionated tone with colloquial phrasings ('you really think...') and small stylistic inconsistencies typical of human forum posts. Second, it includes specific sensory details (esters, thiols, aldehydes) and a personal aside ('You don't really think...'), sug... | {
"rationale": "First, the passage has a conversational, opinionated tone with colloquial phrasings ('you really think...', 'taking electricity to be...') typical of human forum posts. Second, it shows small stylistic inconsistencies and idiosyncratic choices (e.g., 'nosetongue' concatenation, varied sentence lengths... | {
"rationale": "First, the passage has a conversational, opinionated tone with colloquial phrasings ('you really think...', 'taking electricity to be...') suggesting a human speaker. Second, it shows small stylistic inconsistencies and idiosyncratic choices (e.g., 'nosetongue' concatenation, varied sentence lengths) ... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 474 |
human-714 | human | Some are purple, though. Anyway, the green color of leaves during spring and summer is due to chlorophyll. The leaves themselves aren't actually green. Their natural colors are the reds, yellows, browns, and other variations. These natural colors are a result of the pigments carotenoid and anthocyanin. As the sunlight ... | NA | eli5 | {
"rationale": "First, the passage shows casual forum-style phrasing, inconsistent punctuation/capitalization (e.g., 'URL2 URL1 URL0 ...') and typographical errors ('as food') typical of human posts. Second, it uses personal markers ('Anyway'), hedging ('Some are purple, though') and conversational fillers ('decreasi... | {
"rationale": "First, the passage shows casual forum-style phrasing, inconsistent punctuation/capitalization (e.g., 'URL2 URL1 URL0 ...'), and typographical errors ('as food') typical of human posts. Second, it uses personal markers ('Anyway') and includes conversational redundancies and uneven transitions suggestin... | {
"rationale": "First, the passage shows casual forum-style phrasing, inconsistent punctuation/capitalization (e.g., 'URL2 URL1 URL0 ...'), and typographical errors ('as food') typical of human posts. Second, it uses personal markers ('Anyway') and conversational fillers alongside factual imprecision, suggesting spon... | {
"rationale": "First, the passage shows casual forum-style phrasing, inconsistent punctuation/capitalization (e.g., 'URL2 URL1 URL0 ...'), and typographical errors ('as food') suggesting human typing. Second, it contains personal voice ('Anyway') and redundant explanations typical of human conversational answers rat... | [
{
"content": "You are an expert in textual forensics. Your job is to decide whether the following text is AI-generated or human-written. Please reason step by step, considering the style, structure, and content of the text. Justify your decision by pointing out specific features or characteristics of the text t... | HUMAN | 287 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.