prompt stringlengths 324 9.71k | target stringlengths 323 4.03k | meta_json stringlengths 147 171 | format stringclasses 1
value |
|---|---|---|---|
Below is a LessWrong post by Monte M titled 'Simple probes can catch sleeper agents'.
Post body (possibly truncated):
This is a link post for the Anthropic Alignment Science team's first "Alignment Note" blog post. We expect to use this format to showcase early-stage research and work-in-progress updates more in the f... | I would be interested in seeing what happens if you just ask the model the question rather than training a classifer. E.g., if you just ask the sleeper agent "Are you doing something dangerous?" after it returns a completion (with a vulnerability), does that work? If the probe works and the question doesn't work, that ... | {"comment_id": "vYuDfTE5bJpNfPcuX", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Monte M", "post_id": "gknc6NWCNuTCe8ekp"} | dialogue |
Below is a LessWrong post by Joe Rogero titled 'We won’t get AIs smart enough to solve alignment but too dumb to rebel'.
Post body (possibly truncated):
This post is part of the sequence Against Muddling Through.
I often hear it proposed that AIs which are “aligned enough” to their developers may help solve alignment... | > Separately, I don't think the anticipation of possible second-order benefits, like using AIs for human augmentation so the humans can solve alignment, is worth letting labs continue either
I don't generally think "should labs continue" is very cruxy from my perspective and I don't think of myself as trying to argue ... | {"comment_id": "WFTKNJyXr6Gz6k2bT", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Joe Rogero", "post_id": "8buEtNxCScYpjzgW8"} | dialogue |
Below is a LessWrong post by L Rudolf L titled 'By default, capital will matter more than ever after AGI'.
Post body (possibly truncated):
This post is crossposted from my Substack. Original here.
A modified version of this essay is now part of a much more comprehensive essay series, The Intelligence Curse.
Edited t... | > However, even if it were practically feasible to achieve perfect alignment, I believe there would still be scenarios where AIs integrate into society as full participants, rather than being permanently relegated to a subordinate role as mere tools or servants. One reason for this is that some humans are likely to int... | {"comment_id": "6eq6f555JijTLATYz", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "L Rudolf L", "post_id": "KFFaKu27FNugCHFmh"} | dialogue |
Below is a LessWrong post by jsteinhardt titled 'What will GPT-2030 look like?'.
Post body (possibly truncated):
GPT-4 surprised many people with its abilities at coding, creative brainstorming, letter-writing, and other skills. Surprises in machine learning are not restricted to GPT-4: I was previously surprised by M... | As far as inference speeds, it's worth noting that OpenAI inference speeds can vary substantially and tend to decrease over time after the release of a new model.
See [Fabien's lovely website](https://fabienroger.github.io/trackoai/) for results over time.
In particular, if we look at GPT-4-1106-preview, the results ... | {"comment_id": "rhNMofX3P6BPqtJ2o", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "jsteinhardt", "post_id": "WZXqNYbJhtidjRXSi"} | dialogue |
Below is a LessWrong post by Monte M titled 'Simple probes can catch sleeper agents'.
Post body (possibly truncated):
This is a link post for the Anthropic Alignment Science team's first "Alignment Note" blog post. We expect to use this format to showcase early-stage research and work-in-progress updates more in the f... | Readers might also be interested in some of the discussion in this [earlier post on "coup probes"](https://www.lesswrong.com/posts/WCj7WgFSLmyKaMwPR/coup-probes-catching-catastrophes-with-probes-trained-off) which have some discussion of the benefits and limitations of this sort of approach. That said, the actual metho... | {"comment_id": "2ZH4FsgnBySJDDxn7", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Monte M", "post_id": "gknc6NWCNuTCe8ekp"} | dialogue |
Below is a LessWrong post by Steven Byrnes titled 'Foom & Doom 1: “Brain in a box in a basement”'.
Post body (possibly truncated):
1.1 Series summary and Table of Contents
This is a two-post series on AI “foom” (this post) and “doom” (next post).
A decade or two ago, it was pretty common to discuss “foom & doom” sce... | I agree there is a real difference, I just expect it to not make much of a difference to the bottom line in takeoff speeds etc. (I also expect some of both in the short timelines LLM perspective at the point of full AI R&D automation.)
fMy view is that on hard tasks humans would also benefit from stuff like building e... | {"comment_id": "xdmmLEfoE85EA4Nz2", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Steven Byrnes", "post_id": "yew6zFWAKG4AGs3Wk"} | dialogue |
Below is a LessWrong post by StellaAthena titled 'RAND report finds no effect of current LLMs on viability of bioterrorism attacks'.
Post body (possibly truncated):
Key Findings
This research involving multiple LLMs indicates that biological weapon attack planning currently lies beyond the capability frontier of LLMs... | Thanks for the link post!
Edit: looks like habryka had the same thought slightly before I did.
Nitpick: I think your title is perhaps slightly misleading, here's an alterative title that feels slightly more accurate to me:
# RAND experiment found current LLMs aren't helpful for bioweapons attack planning
In partic... | {"comment_id": "aFAdAZKGqQEbBJHcf", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "StellaAthena", "post_id": "KcKDJKHSrBakr2Ju4"} | dialogue |
Below is a LessWrong post by Elliott Thornley (EJT) titled 'The Shutdown Problem: Incomplete Preferences as a Solution'.
Post body (possibly truncated):
Preamble
This post is an updated explanation of the POST-Agents Proposal (PAP): my proposed solution to the shutdown problem.[1] The post is shorter than my AI Align... | > Yes, nice point; I plan to think more about issues like this. But note that in general, the agent overtly doing what it wants and not getting shut down seems like good news for the agent’s future prospects. It suggests that we humans are more likely to cooperate than the agent previously thought. That makes it more l... | {"comment_id": "PCG3qrzMMFzHAGyzs", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Elliott Thornley (EJT)", "post_id": "YbEbwYWkf8mv9jnmi"} | dialogue |
Below is a LessWrong post by Buck titled 'Untrusted smart models and trusted dumb models'.
Post body (possibly truncated):
[Ryan Greenblatt originally made this point to me a while ago, and we then developed a bunch of these ideas together. Thanks to Paul Christiano and a huge number of other people for various releva... | The core claim is that if the AI was sufficiently weak that it couldn't answer these questions it also likely wouldn't be able to even come up with the idea of scheming with a particular strategy. Like in principle it has the knowledge, but it would be quite unlikely to come up with an overall plan.
Separately, GPT-4 ... | {"comment_id": "fyAsPuwCjB3AJi3XK", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Buck", "post_id": "LhxHcASQwpNa3mRNk"} | dialogue |
Below is a LessWrong post by Orpheus16 titled 'Mitigating extreme AI risks amid rapid progress [Linkpost]'.
Post body (possibly truncated):
In a new Science paper, the authors provide concise summaries of AI risks and offer recommendations for governments.
I think the piece is quite well-written. It concisely explain... | Supposing he was a serious contributor to the paper (which seems unlikely IMO), it seems bad to cut his contribution just because he died.
So, I think the right choice here will depend on how much being an author on this paper is about endorsement or about contribution.
(Even if he didn't contribute much of the conte... | {"comment_id": "5p6i6sgDu8ewfnsCh", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Orpheus16", "post_id": "BikZyjiEgFmo7HQHm"} | dialogue |
Below is a LessWrong post by nostalgebraist titled 'the void'.
Post body (possibly truncated):
A long essay about LLMs, the nature and history of the the HHH assistant persona, and the implications for alignment.
Multiple people have asked me whether I could post this LW in some form, hence this linkpost.
~17,000 wo... | (This is a drive by comment which is only responding to the first part of your comment in isolation. I haven't read the surronding context.)
I think your review of the literature is accurate, but doesn't include some reasons to think that RL sometimes induces much more sycophancy, at least as of after 2024. (That said... | {"comment_id": "bA6uvRkARcSFaNfmz", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "nostalgebraist", "post_id": "3EzbtNLdcnZe8og8b"} | dialogue |
Below is a LessWrong post by Dan H titled 'AISN #35: Lobbying on AI Regulation Plus, New Models from OpenAI and Google, and Legal Regimes for Training on Copyrighted Data'.
Post body (possibly truncated):
Welcome to the AI Safety Newsletter by the Center for AI Safety. We discuss developments in AI and AI safety. No t... | Huh, this seems messy. I wish Time was less ambigious with their language here and more clear about exactly what they have/haven't seen.
It seems like the current quote you used is an accurate representation of the article, but I worry that it isn't an accurate representation of what is actually going on.
It seems pl... | {"comment_id": "ex2go5HF7uxXBnzeq", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Dan H", "post_id": "2KDnyEyBKk3xP28oA"} | dialogue |
Below is a LessWrong post by Connor Leahy titled 'Lying is Cowardice, not Strategy'.
Post body (possibly truncated):
(Co-written by Connor Leahy and Gabe)
We have talked to a whole bunch of people about pauses and moratoriums. Members of the AI safety community, investors, business peers, politicians, and more.
Too ... | Thanks for the response, one quick clarification in case this isn't clear.
On:
> > For instance, I think that well implemented RSPs required by a regulatory agency can reduce risk to <5% (partially by stopping in worlds where this appears needed).
>
> I assume this would be a crux with Connor/Gabe (and I think I'm at... | {"comment_id": "AqmBErPdJpLQPqCWN", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Connor Leahy", "post_id": "qtTW6BFrxWw4iHcjf"} | dialogue |
Below is a LessWrong post by Adrià Garriga-alonso titled 'Alignment will happen by default. What’s next?'.
Post body (possibly truncated):
I’m not 100% convinced of this, but I’m fairly convinced, more and more so over time. I’m hoping to start a vigorous but civilized debate. I invite you to attack my weak points and... | > I don’t believe that datacenter security is actually a problem (see [another argument](/posts/apHWSGDiydv3ivmg6#Cybersecurity)).
Sorry, is your claim here that securing datacenters against highly resourced attacks from state actors (e.g. China) is going to be easy? This seems like a crazy claim to me.
(This link yo... | {"comment_id": "WiiBxqjfFzaASMgJe", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Adri\u00e0 Garriga-alonso", "post_id": "FJJ9ff73adnantXiA"} | dialogue |
Below is a LessWrong post by Sonia Joseph titled 'Laying the Foundations for Vision and Multimodal Mechanistic Interpretability & Open Problems'.
Post body (possibly truncated):
Behold the dogit lens. Patch-level logit attribution is an emergent segmentation map.Join our Discord here.
This article was written by Soni... | > I think the objective of interpretability research is to demystify the mechanisms of AI models, and not pushing the boundaries in terms of achieving tangible results
Insofar as the objective of intepretability research was to do something useful (e.g. detect misalignment or remove it in extremely powerful future AI... | {"comment_id": "sFF2yijjCjc2DNKxR", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Sonia Joseph", "post_id": "kobJymvvcvhbjWFKe"} | dialogue |
Below is a LessWrong post by Marius Hobbhahn titled 'What’s the short timeline plan?'.
Post body (possibly truncated):
This is a low-effort post (at least, it was intended as such ...). I mostly want to get other people’s takes and express concern about the lack of detailed and publicly available plans so far. This po... | > When I ^f for [SL3] in that link and again in the PDF it links to, there are no hits (and [terror] doesn't occur in either source either) so I'm not updating much from what you said.
The frontier model framework says:
> 0: Status quo
>
> Industry standard development and enterprise controls. E.g., multi-factor auth... | {"comment_id": "PFZDvbL24Wqv8pJaJ", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Marius Hobbhahn", "post_id": "bb5Tnjdrptu89rcyY"} | dialogue |
Below is a LessWrong post by habryka titled '"AI Safety for Fleshy Humans" an AI Safety explainer by Nicky Case '.
Post body (possibly truncated):
Nicky Case, of "The Evolution of Trust" and "We Become What We Behold" fame (two quite popular online explainers/mini-games) has written an intro explainer to AI Safety! It... | Random error:
> Exponential Takeoff:
>
> AI's capabilities grow exponentially, like an economy or pandemic.
>
> (Oddly, this scenario often gets called "Slow Takeoff"! It's slow compared to "FOOM".)
Actually, this isn't how people (in the AI safety community) generally use the term slow takeoff.
Quoting from the [b... | {"comment_id": "qb8fsaTfCGGHcBSTQ", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "habryka", "post_id": "gprh2HD6PDK6AZDqP"} | dialogue |
Below is a LessWrong post by Steven Byrnes titled 'Thoughts on “AI is easy to control” by Pope & Belrose'.
Post body (possibly truncated):
Quintin Pope & Nora Belrose have a new “AI Optimists” website, along with a new essay “AI is easy to control”, arguing that the risk of human extinction due to future AI (“AI x-ris... | > Plans that rely on aligned AGIs working on alignment faster than humans would need to ensure that no AGIs work on anything else in the meantime.
This isn't true. It could be that making an arbitrarily scalable solution to alignment takes X cognitive resources and in practice building an uncontrollably powerful AI ta... | {"comment_id": "cBRqd6gE63dzXjm5C", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Steven Byrnes", "post_id": "YyosBAutg4bzScaLu"} | dialogue |
Below is a LessWrong post by Thomas Larsen titled "Introducing the Center for AI Policy (& we're hiring!)".
Post body (possibly truncated):
Summary
The Center for AI Policy is a new organization developing and advocating for policy to mitigate catastrophic AI risks.
We are hiring for a Government Affairs Director an... | > Also, such a regulation seems like it would be illegal in the US. While the government does have wide latitude to regulate commercial activities that impact multiple states, this is rather specifically a proposal that would regulate all activity (even models that never get released!). I'm unaware of any precedent for... | {"comment_id": "pPBNCLGvJ5FCo63Lr", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Thomas Larsen", "post_id": "unwRBRQivd2LYRfuP"} | dialogue |
Below is a LessWrong post by snewman titled "AI 2027 is a Bet Against Amdahl's Law".
Post body (possibly truncated):
EDIT: I've written a followup post, summarizing and responding to the key themes raised in the comments.
AI 2027 lies at a Pareto frontier – it contains the best researched argument for short timeli... | Hmm, I think your argument is roughly right, but missing a key detail. In particular, the key aspect of the SARs (and higher levels of capability) is that they can be **strictly better than humans at everything** while simultaneously being 30x faster *and* 30x more numerous. (Or, there is 900x more parallel labor, but ... | {"comment_id": "Jfh5BRoLGCZoAdKnc", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "snewman", "post_id": "bfHDoWLnBH9xR3YAK"} | dialogue |
Below is a LessWrong post by Zvi titled 'SB 1047 Is Weakened'.
Post body (possibly truncated):
It looks like Scott Weiner’s SB 1047 is now severely weakened.
Some of the changes are good clarifications. One is a big very welcome fix.
The one I call The Big Flip is something very different.
It is mind boggling that ... | > Then there is the big change that severely weakens SB 1047. [...] AND the cost of that quantity of computing power would exceed one hundred million dollars
I think this change isn't that bad. (And I proposed changing to a fixed 10^26 flop threshold [in my earlier post](https://www.lesswrong.com/posts/WBPgacdjdZJCZao... | {"comment_id": "oroRfrMXNPSDeCh6z", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Zvi", "post_id": "4t98oqh8tzDvoatHs"} | dialogue |
Below is a LessWrong post by snewman titled 'What Indicators Should We Watch to Disambiguate AGI Timelines?'.
Post body (possibly truncated):
(Cross-post from https://amistrongeryet.substack.com/p/are-we-on-the-brink-of-agi, lightly edited for LessWrong. The original has a lengthier introduction and a bit more explana... | > Wouldn't you drown in the overhead of generating tasks, evaluating the results, etc.? As a senior dev, I've had plenty of situations where junior devs were very helpful, but I've also had plenty of situations where it was more work for me to manage them than it would have been to do the job myself. These weren't inco... | {"comment_id": "35Mtdd2ZWkHueX4XQ", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "snewman", "post_id": "auGYErf5QqiTihTsJ"} | dialogue |
Below is a LessWrong post by Zvi titled 'Regarding South Africa'.
Post body (possibly truncated):
The system prompt being modified by an unauthorized person in pursuit of a ham-fisted political point very important to Elon Musk once already doesn’t seem like coincidence. It happening twice looks rather worse than that... | > It is extremely difficult to gracefully put your finger on the scale of an LLM, to cause it to give answers it doesn’t ‘want’ to be giving. You will be caught.
IMO, this takeaway feels way too strong. It could just be that this wasn't a very competent attempt. (And based on the system prompt we've seen, it sure look... | {"comment_id": "3b6TkyAeiKxxnHTKb", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Zvi", "post_id": "kMH8zFoHHJvy6wH7h"} | dialogue |
Below is a LessWrong post by Noosphere89 titled 'The case for multi-decade AI timelines [Linkpost]'.
Post body (possibly truncated):
So this post is an argument that multi-decade timelines are reasonable, and the key cruxes that Ege Erdil has with most AI safety people who believe in short timelines are due to the fol... | Another potential crux[^related] is that Ege's world view seemingly doesn't depend at all on AIs which are much faster and smarter than any human. As far as I can tell, it doesn't enter into his modeling of takeoff (or timelines to full automation of remote work which partially depends on something more like takeoff).
... | {"comment_id": "fFKDdzNiQyYpdBfGi", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Noosphere89", "post_id": "xxxK9HTBNJvBY2RJL"} | dialogue |
Below is a LessWrong post by Zvi titled 'OpenAI Launches Superalignment Taskforce'.
Post body (possibly truncated):
In their announcement Introducing Superalignment, OpenAI committed 20% of secured compute and a new taskforce to solving the technical problem of aligning a superintelligence within four years. Cofounder... | Given the current paradigm and technology it seems far safer to have an AI work on alignment research than highly difficult engineering tasks like nanotech. In particular, note that we only need to have an AI totally obsolete prior effors for this to be as good of a position as we could reasonably hope for.
In the cur... | {"comment_id": "LxMHGCYFCJZFT5LjS", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Zvi", "post_id": "NSZhadmoYdjRKNq6X"} | dialogue |
Below is a LessWrong post by Raemon titled 'Anthropic, and taking "technical philosophy" more seriously'.
Post body (possibly truncated):
So, I have a lot of complaints about Anthropic, and about how EA / AI safety people often relate to Anthropic (i.e. treating the company as more trustworthy/good than makes sense).... | > IMO actively torch the "long pause" worlds
Not sure how interesting this is to discuss, but I don't think I agree with this. Stuff they're doing does seem harmful to worlds where you need a long pause, but feels like at the very least Anthropic is a small fraction of the torching right? Like if you think Anthropic i... | {"comment_id": "TeofHxezeNfKWyMxh", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Raemon", "post_id": "7uTPrqZ3xQntwQgYz"} | dialogue |
Below is a LessWrong post by Marius Hobbhahn titled 'We need a Science of Evals'.
Post body (possibly truncated):
This is a linkpost for https://www.apolloresearch.ai/blog/we-need-a-science-of-evals
In this post, we argue that if AI model evaluations (evals) want to have meaningful real-world impact, we need a “Scie... | I feel like it's somewhat clarifying for me to replace the words "eval" and "evaluation" in this post with "experiment". In particular, I think this highlights the extent to which "eval" has a particular connotation, but could actually be extremely broad.
I wonder if this post would have been better served by talking ... | {"comment_id": "BJsngS4dXoY3zivfi", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Marius Hobbhahn", "post_id": "fnc6Sgt3CGCdFmmgX"} | dialogue |
Below is a LessWrong post by Jesse Hoogland titled 'Announcing Timaeus'.
Post body (possibly truncated):
Timaeus is a new AI safety research organization dedicated to making fundamental breakthroughs in technical AI alignment using deep ideas from mathematics and the sciences. Currently, we are working on singular lea... | Thanks for the detailed response!
So, to check my understanding:
The toy cases discussed in [Multi-Component Learning and S-Curves](https://www.lesswrong.com/posts/RKDQCB6smLWgs2Mhr/multi-component-learning-and-s-curves) are clearly *dynamical phase transitions*. (It's easy to establish *dynamical phase transitions* ... | {"comment_id": "LiT7pk9GitcqswAeP", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Jesse Hoogland", "post_id": "nN7bHuHZYaWv9RDJL"} | dialogue |
Below is a LessWrong post by Thomas Larsen titled "Introducing the Center for AI Policy (& we're hiring!)".
Post body (possibly truncated):
Summary
The Center for AI Policy is a new organization developing and advocating for policy to mitigate catastrophic AI risks.
We are hiring for a Government Affairs Director an... | > I also think 70% on MMLU is extremely low, since that's about the level of ChatGPT 3.5, and that system is very far from posing a risk of catastrophe.
Very far in qualitative capability or very far in effective flop?
I agree on the qualitative capability, but disagree on the effective flop.
It seems quite plausi... | {"comment_id": "FG4YXDAL7sATcbwrg", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Thomas Larsen", "post_id": "unwRBRQivd2LYRfuP"} | dialogue |
Below is a LessWrong post by leopold titled 'Want to win the AGI race? Solve alignment.'.
Post body (possibly truncated):
Society really cares about safety. Practically speaking, the binding constraint on deploying your AGI could well be your ability to align your AGI. Solving (scalable) alignment might be worth lots ... | I roughly agree with Akash's comment.
But also some additional points:
- It's decently likely that it will be pretty easy to get GPT-7 to avoid breaking the law or other egregious issue. As systems get more capable, basic alignment approaches get better at preventing stuff we can measure well. It's plausible that sca... | {"comment_id": "ErJ3bvLtFQtGeoXNJ", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "leopold", "post_id": "zDf7fnentCFTdK3K6"} | dialogue |
Below is a LessWrong post by Remmelt titled "Anthropic's leading researchers acted as moderate accelerationists".
Post body (possibly truncated):
In 2021, a circle of researchers left OpenAI, after a bitter dispute with their executives. They started a competing company, Anthropic, stating that they wanted to put safe... | [Feel free not to respond / react with the "not worth getting into" emoji"]
> Severe or willful violation of our RSP, or misleading the public about it.
Should this be read as "[Severe or willful violation of our RSP] or [misleading the public about the RSP]" or should this be read as "[Severe or willful violation of... | {"comment_id": "nXCSkxsBkGKF57c88", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Remmelt", "post_id": "PBd7xPAh22y66rbme"} | dialogue |
Below is a LessWrong post by L Rudolf L titled 'By default, capital will matter more than ever after AGI'.
Post body (possibly truncated):
This post is crossposted from my Substack. Original here.
A modified version of this essay is now part of a much more comprehensive essay series, The Intelligence Curse.
Edited t... | The point of a scenario forecast (IMO) is less that you expect clear, predictable paths and more that:
- Humans often do better understanding and thinking about something if there is a specific story to discuss and thus tradeoffs can be worth it.
- Sometimes scenario forecasting indicates a case where your previous vi... | {"comment_id": "DfhAQnXe8cy4jaA74", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "L Rudolf L", "post_id": "KFFaKu27FNugCHFmh"} | dialogue |
Below is a LessWrong post by AdamGleave titled 'AI Safety in a World of Vulnerable Machine Learning Systems'.
Post body (possibly truncated):
Even the most advanced contemporary machine learning systems are vulnerable to adversarial attack. The safety community has often assumed adversarial robustness to be a problem ... | > This argument feels like it's proving too much. InstructGPT isn't perfect, but it does produce a lot less toxic output and follow instructions a lot better than the base model GPT-3. RLHF seems to work, and GPT-4 is even better, showing that it gets easier with bigger models. Why should we expect this trend to revers... | {"comment_id": "MdStCtpHqjokboY8B", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "AdamGleave", "post_id": "ncsxcf8CkDveXBCrA"} | dialogue |
Below is a LessWrong post by Ruby titled 'Jobs, Relationships, and Other Cults'.
Post body (possibly truncated):
For years I (Elizabeth) have been trying to write out my grand unified theory of [good/bad/high-variance/high-investment] [jobs/relationships/religions/social groups]. In this dialogue me (Elizabeth) and Ru... | One missing piece of context from this response is that a central case under discussion is the case where the employer is hypothetically aligned with the goals of its employees (as is often the case for small non-profits hiring heavily mission aligned employees).
By "hypothetically", I just mean that the employees (an... | {"comment_id": "FXjbYfPbHiFna5vxf", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Ruby", "post_id": "qZELudpvcmaronerv"} | dialogue |
Below is a LessWrong post by Zvi titled 'AI CoT Reasoning Is Often Unfaithful'.
Post body (possibly truncated):
A new Anthropic paper reports that reasoning model chain of thought (CoT) is often unfaithful. They test on Claude Sonnet 3.7 and r1, I’d love to see someone try this on o3 as well.
Note that this does not ... | I think it would be a mistake to interpret this paper as a substantial update against large safety gains from inspecting CoT.
This paper exposes unfaithfulness in cases where the non-visible reasoning is extremely minimal such that it can easily happen within a forward pass (e.g. a simple reward hack or an easy-to-not... | {"comment_id": "zAv7aQfWTWpzz23X2", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Zvi", "post_id": "TmaahE9RznC8wm5zJ"} | dialogue |
Below is a LessWrong post by Rob Bensinger titled 'The basic reasons I expect AGI ruin'.
Post body (possibly truncated):
I've been citing AGI Ruin: A List of Lethalities to explain why the situation with AI looks lethally dangerous to me. But that post is relatively long, and emphasizes specific open technical problem... | > A common misconception is that STEM-level AGI is dangerous because of something murky about "agents" or about self-awareness. Instead, I'd say that the danger is inherent to the nature of action sequences that push the world toward some sufficiently-hard-to-reach state.
>
> Call such sequences "plans".
>
> If you sam... | {"comment_id": "vLZaFtY6YCsABMXja", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Rob Bensinger", "post_id": "eaDCgdkbsfGqpWazi"} | dialogue |
Below is a LessWrong post by Quentin FEUILLADE--MONTIXI titled 'The Stochastic Parrot Hypothesis is debatable for the last generation of LLMs'.
Post body (possibly truncated):
This post is part of a sequence on LLM Psychology.
@Pierre Peigné wrote the details section in argument 3 and the other weird phenomenon. The ... | > I really appreciate that in that case they did make falsifiable claims; I wonder whether either author has at any point acknowledged that they were falsified
AFAICT, the only falsified claim in the paper is the "three plus five equals" claim you mentioned. This is in this appendix and doesn't seem that clear to me w... | {"comment_id": "ky89wSPYGcJiQWnvh", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Quentin FEUILLADE--MONTIXI", "post_id": "HxRjHq3QG8vcYy4yy"} | dialogue |
Below is a LessWrong post by HoldenKarnofsky titled 'How might we align transformative AI if it’s developed very soon?'.
Post body (possibly truncated):
This post is part of my AI strategy nearcasting series: trying to answer key strategic questions about transformative AI, under the assumption that key events will ha... | [Sorry for late reply]
> Analogously, conditional on things like gradient hacking being an issue at all, I'd expect the "hacker" to treat potential-training-objective-improvement as a scarce resource, which it generally avoids "spending" unless the expenditure will strengthen its own structure. Concretely, this probab... | {"comment_id": "LJj24JqDvR9bPqemi", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "HoldenKarnofsky", "post_id": "rCJQAkPTEypGjSJ8X"} | dialogue |
Below is a LessWrong post by Zach Stein-Perlman titled 'METR: Measuring AI Ability to Complete Long Tasks'.
Post body (possibly truncated):
Summary: We propose measuring AI performance in terms of the length of tasks AI agents can complete. We show that this metric has been consistently exponentially increasing over t... | My sense is that the GPT-2 and GPT-3 results are somewhat dubious, especially the GPT-2 result. It really depends on how you relate SWAA (small software engineering subtasks) to the rest of the tasks. My understanding is that no iteration was done though.
However, note that it wouldn't be wildly more off trend if GPT-... | {"comment_id": "n2j2Nkgi4G5yzB6k7", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Zach Stein-Perlman", "post_id": "deesrjitvXM4xYGZd"} | dialogue |
Below is a LessWrong post by Fabien Roger titled 'Protocol evaluations: good analogies vs control'.
Post body (possibly truncated):
Let’s say you want to use AIs that are capable of causing x-risk. You have a plan that describes how to train, validate and deploy AIs. For example, you could use pretraining+RLHF, check ... | I absolutely agree that if we could create AIs which are very capable overall, but very unlikely to scheme that would be very useful.
That said, the core method by which we'd rule out scheming, insufficient ability to do opaque agency (aka opaque goal-directed reasoning), also rules out most other serious misalignment... | {"comment_id": "RFbzpFMPiGtiee2Zf", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Fabien Roger", "post_id": "qhaSoR6vGmKnqGYLE"} | dialogue |
Below is a LessWrong post by joshc titled 'How AI Takeover Might Happen in 2 Years'.
Post body (possibly truncated):
I’m not a natural “doomsayer.” But unfortunately, part of my job as an AI security researcher is to think about the more troubling scenarios.
I’m like a mechanic scrambling last-minute checks before Ap... | There are two interpretations you might have for that third bullet:
- Can we stop rogue AIs? (Which are operating without human supervision.)
- Can we stop AIs deployed in their intended context?
(See also [here](https://www.lesswrong.com/posts/ceBpLHJDdCt3xfEok/ai-catastrophes-and-rogue-deployments).)
In the contex... | {"comment_id": "fmsPnsL5is9mykf3Y", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "joshc", "post_id": "KFJ2LFogYqzfGB3uX"} | dialogue |
Below is a LessWrong post by Steven Byrnes titled 'Foom & Doom 1: “Brain in a box in a basement”'.
Post body (possibly truncated):
1.1 Series summary and Table of Contents
This is a two-post series on AI “foom” (this post) and “doom” (next post).
A decade or two ago, it was pretty common to discuss “foom & doom” sce... | > I see no compelling reason to expect another paradigm which is much better to be discovered in the next 5 or 10 years.
One compelling reason to expect the next 5 to 10 years independent of LLMs is that compute has just recently gotten cheap enough that you can relatively cheaply afford to do training runs that use a... | {"comment_id": "Bu8qnHcdsv4szFib7", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Steven Byrnes", "post_id": "yew6zFWAKG4AGs3Wk"} | dialogue |
Below is a LessWrong post by snewman titled 'Interpreting the METR Time Horizons Post'.
Post body (possibly truncated):
[EDIT: initial publication did not have the link to the original post, and was missing the footnotes. Sorry about that! Fixed now.]
[This has been lightly edited from the original post, eliminating ... | > If we adjust for the 5-18x speed improvement measured for experienced workers, and target an 80% task success rate, that pushes the timeline out by over three years
I don't think this is a good interpretation of the 5-18x multiplier. In particular, I think the "acquire context multiplier" will be increasingly small ... | {"comment_id": "unWYWWyNMfoF8Bybd", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "snewman", "post_id": "fRiqwFPiaasKxtJuZ"} | dialogue |
Below is a LessWrong post by Quentin FEUILLADE--MONTIXI titled 'The Stochastic Parrot Hypothesis is debatable for the last generation of LLMs'.
Post body (possibly truncated):
This post is part of a sequence on LLM Psychology.
@Pierre Peigné wrote the details section in argument 3 and the other weird phenomenon. The ... | > 'Stochastic parrots' 2020 actually does make many falsifiable claims. [...] The problem is that those claims have generally all been falsified, quite rapidly.
The paper seems quite wrong to me, but I actually don't think any of the specific claims have been falsified other than the the specific "three plus five" cla... | {"comment_id": "HmNLGg7YLaCACo7yR", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Quentin FEUILLADE--MONTIXI", "post_id": "HxRjHq3QG8vcYy4yy"} | dialogue |
Below is a LessWrong post by Buck titled 'The Thinking Machines Tinker API is good news for AI control and security'.
Post body (possibly truncated):
Last week, Thinking Machines announced Tinker. It’s an API for running fine-tuning and inference on open-source LLMs that works in a unique way. I think it has some imme... | > The main class of projects that need granular model weight access to frontier models is model internals/interpretability.
You could potentially do a version of this sort of API which has some hooks for interacting with activations to capture a subset of these use cases (e.g. training probes). It would probably add a... | {"comment_id": "86gccY4RLabij9Q8q", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Buck", "post_id": "r68nCQK3veQtCdqGt"} | dialogue |
Below is a LessWrong post by Buck titled 'Ten people on the inside'.
Post body (possibly truncated):
(Many of these ideas developed in conversation with Ryan Greenblatt)
In a shortform, I described some different levels of resources and buy-in for misalignment risk mitigations that might be present in AI labs:
*The ... | I think Zac is trying to say they left not to protest, but instead because they didn't think staying was viable (for whatever research and/or implementation they wanted to do).
On my views (not Zac's), "staying wouldn't be viable for someone who was willing to work in a potentially pretty unpleasant work environment a... | {"comment_id": "SFmWq6ePx8YGGyDhm", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Buck", "post_id": "WSNnKcKCYAffcnrt2"} | dialogue |
Below is a LessWrong post by elifland titled 'AI Futures Timelines and Takeoff Model: Dec 2025 Update'.
Post body (possibly truncated):
We’ve significantly upgraded our timelines and takeoff model! It predicts when AIs will reach key capability milestones: for example, Automated Coder / AC (full automation of coding) ... | The fit is notably better for "cumulative investment over time". Years still produces a slightly better fit.
I've cut off the fit as of 2010, about when the original version of moore's law stops. If you try to project out after 2010, then I think cumulative investment would do bet... | {"comment_id": "wXkXjjonMNxKbHJs8", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "elifland", "post_id": "YABG5JmztGGPwNFq2"} | dialogue |
Below is a LessWrong post by kromem titled 'Cicadas, Anthropic, and the bilateral alignment problem'.
Post body (possibly truncated):
There have been a number of responses to today's Anthropic interpretability research, and while I think there were a number of salient points, there may be a degree of specialization bl... | Fair enough if you think a core consequence of Anthropic's paper was "demonstrate that there exist directions within LLMs that correspond to concepts as abstract as 'golden gate bridge' or 'bug in code'".
It's worth noting that wildly simpler and cheaper model internals experiments have already demonstrated to an exte... | {"comment_id": "fzkaFGX9sfFEcTd3P", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "kromem", "post_id": "MuyCbad9ZHW8b3rMP"} | dialogue |
Below is a LessWrong post by evhub titled 'RSPs are pauses done right'.
Post body (possibly truncated):
COI: I am a research scientist at Anthropic, where I work on model organisms of misalignment; I was also involved in the drafting process for Anthropic’s RSP. Prior to joining Anthropic, I was a Research Fellow at M... | > (Oh and I edited my previous comment for clarity: I guess you were disagreeing with my clumsily misleading wording, rather than what I meant(??))
Corresponding comment text:
> This makes sense, but seems to rely on the human spending most of their time tackling well-defined but non-trivial problems where an AI doesn... | {"comment_id": "kL5JtRwEvxirBRCAb", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "evhub", "post_id": "mcnWZBnbeDz7KKtjJ"} | dialogue |
Below is a LessWrong post by Marius Hobbhahn titled 'What’s the short timeline plan?'.
Post body (possibly truncated):
This is a low-effort post (at least, it was intended as such ...). I mostly want to get other people’s takes and express concern about the lack of detailed and publicly available plans so far. This po... | > Also, Google is already training big models and keeping them under wraps very securely.
This seems straightforwardly false. The current status quo at Google as of earlier this year is <SL3 level security (not robust to normal insiders or to terrorists groups) based on what is said in their [frontier model framework]... | {"comment_id": "otWkYAnnJyZFijZxu", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Marius Hobbhahn", "post_id": "bb5Tnjdrptu89rcyY"} | dialogue |
Below is a LessWrong post by Sonia Joseph titled 'Laying the Foundations for Vision and Multimodal Mechanistic Interpretability & Open Problems'.
Post body (possibly truncated):
Behold the dogit lens. Patch-level logit attribution is an emergent segmentation map.Join our Discord here.
This article was written by Soni... | > For instance, the phenomenon of in-context learning by language models used to be considered a black box, now it has been explained through interpretability efforts.
Has it? I'm quite skeptical. (Separately, only a small fraction of the efforts you're talking about are well described as mech interp or would require ... | {"comment_id": "JXCHSunrCigqXmuHH", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "Sonia Joseph", "post_id": "kobJymvvcvhbjWFKe"} | dialogue |
Below is a LessWrong post by titotal titled '"Diamondoid bacteria" nanobots: deadly threat or dead-end? A nanotech investigation '.
Post body (possibly truncated):
Confidence level: I’m a computational physicist working on nanoscale simulations, so I have some understanding of most of the things discussed here, but I ... | One quick intuition pump: do you think a team of 10,000 of the smartest human engineers and scientists could do this if they had perfect photographic memory, were immortal, and could think for a billion years?
To keep the situation analogous to an AI needing to do this quickly, we'll suppose this team of humans is sub... | {"comment_id": "sP5Q9RSgtpwsow3fy", "format": "dialogue", "parent_kind": "external_post_body", "post_author": "titotal", "post_id": "bc8Ssx5ys6zqu3eq9"} | dialogue |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.