LessWrong (Curated & Popular)

Latest Episodes:

"When can we trust model evaluations?" bu evhub Aug 09, 2023

Show notes

In "Towards understanding-based safety evaluations," I discussed why I think evaluating specifically the alignment of models is likely to require mechanistic, understanding-based evaluations rather than solely behavioral evaluations. However, I also mentioned in a footnote why I thought behavioral evaluations would likely be fine in the case of evaluating capabilities rather than evaluating alignment:

However, while I like the sorts of behavioral evaluations discussed in the GPT-4 System Card (e.g. ARC's autonomous replication evaluation) as a way of assessing model capabilities, I have a pretty fundamental concern with these sorts of techniques as a mechanism for eventually assessing alignment.

That's because while I think it would be quite tricky for a deceptively aligned AI to sandbag its capabilities when explicitly fine-tuned on some capabilities task (that probably requires pretty advanced gradient hacking), it should be quite easy for such a model to pretend to look aligned.

In this post, I want to try to expand a bit on this point and explain exactly what assumptions I think are necessary for various different evaluations to be reliable and trustworthy. For that purpose, I'm going to talk about four different categories of evaluations and what assumptions I think are needed to make each one go through.
Source:
https://www.lesswrong.com/posts/dBmfb76zx6wjPsBC7/when-can-we-trust-model-evaluations
Narrated for LessWrong by TYPE III AUDIO.
Share feedback on this narration.
[Curated Post] ✓

"ARC Evals new report: Evaluating Language-Model Agents on Realistic Autonomous Tasks" by Beth Barnes Aug 04, 2023

Show notes

Blogpost version

Paper
We have just released our first public report. It introduces methodology for assessing the capacity of LLM agents to acquire resources, create copies of themselves, and adapt to novel challenges they encounter in the wild.

Background

ARC Evals develops methods for evaluating the safety of large language models (LLMs) in order to provide early warnings of models with dangerous capabilities. We have public partnerships with Anthropic and OpenAI to evaluate their AI systems, and are exploring other partnerships as well.
Source:
https://www.lesswrong.com/posts/EPLk8QxETC5FEhoxK/arc-evals-new-report-evaluating-language-model-agents-on
Narrated for LessWrong by TYPE III AUDIO.
Share feedback on this narration.
[125+ Karma Post] ✓

"The "public debate" about AI is confusing for the general public and for policymakers because it is a three-sided debate" by Adam David Long Aug 04, 2023

Show notes

Summary of Argument: The public debate among AI experts is confusing because there are, to a first approximation, three sides, not two sides to the debate. I refer to this as a 🔺three-sided framework, and I argue that using this three-sided framework will help clarify the debate (more precisely, debates) for the general public and for policy-makers.
Source:
https://www.lesswrong.com/posts/BTcEzXYoDrWzkLLrQ/the-public-debate-about-ai-is-confusing-for-the-general
Narrated for LessWrong by TYPE III AUDIO.
Share feedback on this narration.
[125+ Karma Post] ✓

"My current LK99 questions" by Eliezer Yudkowsky Aug 04, 2023

Show notes

So this morning I thought to myself, "Okay, now I will actually try to study the LK99 question, instead of betting based on nontechnical priors and market sentiment reckoning." (My initial entry into the affray, having been driven by people online presenting as confidently YES when the prediction markets were not confidently YES.) And then I thought to myself, "This LK99 issue seems complicated enough that it'd be worth doing an actual Bayesian calculation on it"--a rare thought; I don't think I've done an actual explicit numerical Bayesian update in at least a year.

In the process of trying to set up an explicit calculation, I realized I felt very unsure about some critically important quantities, to the point where it no longer seemed worth trying to do the calculation with numbers. This is the System Working As Intended.
Source:
https://www.lesswrong.com/posts/EzSH9698DhBsXAcYY/my-current-lk99-questions
Narrated for LessWrong by TYPE III AUDIO.
Share feedback on this narration.
[125+ Karma Post] ✓
[Curated Post] ✓

"Thoughts on sharing information about language model capabilities" by paulfchristiano Aug 01, 2023

Show notes

I believe that sharing information about the capabilities and limits of existing ML systems, and especially language model agents, significantly reduces risks from powerful AI—despite the fact that such information may increase the amount or quality of investment in ML generally (or in LM agents in particular).

Concretely, I mean to include information like: tasks and evaluation frameworks for LM agents, the results of evaluations of particular agents, discussions of the qualitative strengths and weaknesses of agents, and information about agent design that may represent small improvements over the state of the art (insofar as that information is hard to decouple from evaluation results).
Source:
https://www.lesswrong.com/posts/fRSj2W4Fjje8rQWm9/thoughts-on-sharing-information-about-language-model
Narrated for LessWrong by TYPE III AUDIO.
Share feedback on this narration.
[125+ Karma Post] ✓
[Curated Post] ✓

"Cultivating a state of mind where new ideas are born" by Henrik Karlsson Jul 31, 2023

Show notes

In the early 2010s, a popular idea was to provide coworking spaces and shared living to people who were building startups. That way the founders would have a thriving social scene of peers to percolate ideas with as they figured out how to build and scale a venture. This was attempted thousands of times by different startup incubators. There are no famous success stories.

In 2015, Sam Altman, who was at the time the president of Y Combinator, a startup accelerator that has helped scale startups collectively worth $600 billion, tweeted in reaction that “not [providing coworking spaces] is part of what makes YC work.” Later, in a 2019 interview with Tyler Cowen, Altman was asked to explain why.

Source:
https://www.lesswrong.com/posts/R5yL6oZxqJfmqnuje/cultivating-a-state-of-mind-where-new-ideas-are-born
Narrated for LessWrong by TYPE III AUDIO.
Share feedback on this narration.
[Curated] ✓

"Self-driving car bets" by paulfchristiano Jul 31, 2023

Show notes

This month I lost a bunch of bets.

Back in early 2016 I bet at even odds that self-driving ride sharing would be available in 10 US cities by July 2023. Then I made similar bets a dozen times because everyone disagreed with me.
Source:
https://www.lesswrong.com/posts/ZRrYsZ626KSEgHv8s/self-driving-car-bets
Narrated for LessWrong by TYPE III AUDIO.
Share feedback on this narration.
[125+ Karma Post] ✓
[Curated Post] ✓

"Yes, It's Subjective, But Why All The Crabs?" by johnswentworth Jul 31, 2023

Show notes

Some early biologist, equipped with knowledge of evolution but not much else, might see all these crabs and expect a common ancestral lineage. That’s the obvious explanation of the similarity, after all: if the crabs descended from a common ancestor, then of course we’d expect them to be pretty similar.

… but then our hypothetical biologist might start to notice surprisingly deep differences between all these crabs. The smoking gun, of course, would come with genetic sequencing: if the crabs’ physiological similarity is achieved by totally different genetic means, or if functionally-irrelevant mutations differ across crab-species by more than mutational noise would induce over the hypothesized evolutionary timescale, then we’d have to conclude that the crabs had different lineages. (In fact, historically, people apparently figured out that crabs have different lineages long before sequencing came along.)

Now, having accepted that the crabs have very different lineages, the differences are basically explained. If the crabs all descended from very different lineages, then of course we’d expect them to be very different.

… but then our hypothetical biologist returns to the original empirical fact: all these crabs sure are very similar in form. If the crabs all descended from totally different lineages, then the convergent form is a huge empirical surprise! The differences between the crab have ceased to be an interesting puzzle - they’re explained - but now the similarities are the interesting puzzle. What caused the convergence?

To summarize: if we imagine that the crabs are all closely related, then any deep differences are a surprising empirical fact, and are the main remaining thing our model needs to explain. But once we accept that the crabs are not closely related, then any convergence/similarity is a surprising empirical fact, and is the main remaining thing our model needs to explain.
Source:
https://www.lesswrong.com/posts/qsRvpEwmgDBNwPHyP/yes-it-s-subjective-but-why-all-the-crabs
Narrated for LessWrong by TYPE III AUDIO.
Share feedback on this narration.
[125+ Karma Post] ✓

"Grant applications and grand narratives" by Elizabeth Jul 27, 2023

Show notes

The Lightspeed application asks: “What impact will [your project] have on the world? What is your project’s goal, how will you know if you’ve achieved it, and what is the path to impact?”

LTFF uses an identical question, and SFF puts it even more strongly (“What is your organization’s plan for improving humanity’s long term prospects for survival and flourishing?”).

I’ve applied to all three grants of these at various points, and I’ve never liked this question. It feels like it wants a grand narrative of an amazing, systemic project that will measurably move the needle on x-risk. But I’m typically applying for narrowly defined projects, like “Give nutrition tests to EA vegans and see if there’s a problem”. I think this was a good project. I think this project is substantially more likely to pay off than underspecified alignment strategy research, and arguably has as good a long tail. But when I look at “What impact will [my project] have on the world?” the project feels small and sad. I feel an urge to make things up, and express far more certainty for far more impact than I believe. Then I want to quit, because lying is bad but listing my true beliefs feels untenable.

I’ve gotten better at this over time, but I know other people with similar feelings, and I suspect it’s a widespread issue (I encourage you to share your experience in the comments so we can start figuring that out).

I should note that the pressure for grand narratives has good points; funders are in fact looking for VC-style megabits. I think that narrow projects are underappreciated, but for purposes of this post that’s beside the point: I think many grantmakers are undercutting their own preferred outcomes by using questions that implicitly push for a grand narrative. I think they should probably change the form, but I also think we applicants can partially solve the problem by changing how we interact with the current forms.

My goal here is to outline the problem, gesture at some possible solutions, and create a space for other people to share data. I didn’t think about my solutions very long, I am undoubtedly missing a bunch and what I do have still needs workshopping, but it’s a place to start.
Source:
https://www.lesswrong.com/posts/FNPXbwKGFvXWZxHGE/grant-applications-and-grand-narratives
Narrated for LessWrong by TYPE III AUDIO.
Share feedback on this narration.
[Curated Post] ✓

"Brain Efficiency Cannell Prize Contest Award Ceremony" by Alexander Gietelink Oldenziel Jul 27, 2023

Show notes

Previously Jacob Cannell wrote the post "Brain Efficiency" which makes several radical claims: that the brain is at the pareto frontier of speed, energy efficiency and memory bandwith, that this represent a fundamental physical frontier.

Here's an AI-generated summary

The article “Brain Efficiency: Much More than You Wanted to Know” on LessWrong discusses the efficiency of physical learning machines. The article explains that there are several interconnected key measures of efficiency for physical learning machines: energy efficiency in ops/J, spatial efficiency in ops/mm^2 or ops/mm^3, speed efficiency in time/delay for key learned tasks, circuit/compute efficiency in size and steps for key low-level algorithmic tasks, and learning/data efficiency in samples/observations/bits required to achieve a level of circuit efficiency, or per unit thereof. The article also explains why brain efficiency matters a great deal for AGI timelines and takeoff speeds, as AGI is implicitly/explicitly defined in terms of brain parity. The article predicts that AGI will consume compute & data in predictable brain-like ways and suggests that AGI will be far more like human simulations/emulations than you’d otherwise expect and will require training/education/raising vaguely like humans1.

Jake further has argued that this has implication for FOOM and DOOM.

Considering the intense technical mastery of nanoelectronics, thermodynamics and neuroscience required to assess the arguments here I concluded that a public debate between experts was called for. This was the start of the Brain Efficiency Prize contest which attracted over a 100 in-depth technically informed comments.

Now for the winners! Please note that the criteria for winning the contest was based on bringing in novel and substantive technical arguments as assesed by me. In contrast, general arguments about the likelihood of FOOM or DOOM while no doubt interesting did not factor into the judgement.

And the winners of the Jake Cannell Brain Efficiency Prize contest are

Ege Erdil
DaemonicSigil
spxtr
... and Steven Byrnes!

Source:
https://www.lesswrong.com/posts/fm88c8SvXvemk3BhW/brain-efficiency-cannell-prize-contest-award-ceremony
Narrated for LessWrong by TYPE III AUDIO.
Share feedback on this narration.
[125+ Karma Post] ✓

Our TOPPODCAST Picks

Follow Us

Stay Connected

LessWrong (Curated & Popular)

Related Podcasts

Links

Stay Connected