text stringlengths 300 320k | source stringlengths 52 154 |
|---|---|
# The King and the Golem
Long ago there was a mighty king who had everything in the world that he wanted, except trust. Who could he trust, when anyone around him might scheme for his throne? So he resolved to study the nature of trust, that he might figure out how to gain it. He asked his subjects to bring him the mo... | https://www.lesswrong.com/posts/bteq4hMW2hqtKE49d/the-king-and-the-golem |
# You should just smile at strangers a lot
This is so basic I feel stupid for saying it in such an intellectual place. But most of you guys share my enthusiasm for effective altruism, and here's something really altruistic and really effective that I've been doing for two decades and I think lots of you should do too... | https://www.lesswrong.com/posts/rYKXgGFvPS7vXCZhP/you-should-just-smile-at-strangers-a-lot |
# What is wrong with this "utility switch button problem" approach?
Suppose you have some positive utility functions $A,B$ , mathematically considered to be random variables dependant on choice of policy.
Let $K$ be a random variable that is 1 if the button is pressed, 0 otherwise. Again dependant on choice of polic... | https://www.lesswrong.com/posts/P8caCHGJdm2GcniAp/what-is-wrong-with-this-utility-switch-button-problem |
# Autonomic Sanity
Epistemic Status: self-reported musings
Defining Sanity
===============
Mental health is a complicated topic, and “sanity” can be a loaded word, so I’ll offer a few definitions to make sure we’re all on the same page.
The colloquial definition of sanity, if such a thing exists, is formed in contr... | https://www.lesswrong.com/posts/DCbyYpv38AfbG5HMa/autonomic-sanity |
# Impact stories for model internals: an exercise for interpretability researchers
*Inspired by Neel's* [*longlist*](https://www.alignmentforum.org/posts/uK6sQCNMw8WKzJeCQ/a-longlist-of-theories-of-impact-for-interpretability)*; thanks to* [*@Nicholas Goldowsky-Dill*](https://www.alignmentforum.org/users/nicholas-gold... | https://www.lesswrong.com/posts/KfDh7FqwmNGExTryT/impact-stories-for-model-internals-an-exercise-for |
# [Linkpost] Mark Zuckerberg confronted about Meta's Llama 2 AI's ability to give users detailed guidance on making anthrax - Business Insider
> Several tech leaders descended upon Capitol Hill last week to discuss the rapid expansion of generative AI. It was a mostly staid meeting until the potential harms from Meta'... | https://www.lesswrong.com/posts/H8NDsEXz6bqusooxj/linkpost-mark-zuckerberg-confronted-about-meta-s-llama-2-ai |
# Making AIs less likely to be spiteful
Which forms of misalignment might result in particularly bad outcomes? And to what extent can we prevent them even if we fail at intent alignment? We define *spite *as a terminal preference for frustrating others’ preferences, at least under some conditions. Reducing the chanc... | https://www.lesswrong.com/posts/92xKPvTHDhoAiRBv9/making-ais-less-likely-to-be-spiteful |
# Announcing the CNN Interpretability Competition
TL;DR
=====
I am excited to announce the [CNN Interpretability Competition,](https://benchmarking-interpretability.csail.mit.edu/challenges-and-prizes/) which is part of the competition track of [SATML 2024](https://satml.org/#).
**Dates:** Sept 22, 2023 - Mar 22, 2... | https://www.lesswrong.com/posts/HXJaKcuQqaZDJq9xz/announcing-the-cnn-interpretability-competition |
# Mechanistic Interpretability Reading group
Greetings.
This is an open invitation for folks interested in technical AI alignment literature to join our reading group, which has been running for 22 weeks now. Our meetings take place online via the *Mechanistic Interpretability Group* Discord server, accessible throug... | https://www.lesswrong.com/posts/ePGwx2uyjnvT8vWEL/mechanistic-interpretability-reading-group |
# ARENA 2.0 - Impact Report
*Authors - Kathryn O'Rourke, Callum McDougall. Thanks to Joseph Bloom and Denizhan Akar for feedback, and major thank you to all of our wonderful TAs and the MATS London team– your positive contributions to the program have been innumerable. *
**TL;DR**
---------
We ran a bootcamp aimed ... | https://www.lesswrong.com/posts/9fbr7axHenRAL5Gkm/arena-2-0-impact-report |
# GPT-4 for personal productivity: online distraction blocker
There are many apps for blocking distracting websites: [freedom.to,](https://freedom.to/) [leechblock](https://www.proginosko.com/leechblock/), [selfcontrol](https://selfcontrolapp.com/), [coldturkey](https://getcoldturkey.com/), just to name a few. They ar... | https://www.lesswrong.com/posts/denTWbbYyTfbYLtE5/gpt-4-for-personal-productivity-online-distraction-blocker |
# Text Posts from the Kids Group: 2022
Another round (previously: [2023](https://www.jefftk.com/p/text-posts-from-the-kids-group-2023-i)) of liberating kid posts from [Facebook](https://www.jefftk.com/p/making-groups-for-kid-pictures):
(Some of these were from me; some were from Julia. Ones saying "me" could mean eit... | https://www.lesswrong.com/posts/g3zYbYP7pePyNchmt/text-posts-from-the-kids-group-2022 |
# Jacob on the Precipice
> And he dreamed, and behold, there was a ladder set up on the earth, and the top of it reached to heaven. And behold, the angels of God were ascending and descending on it! And behold, the LORD stood above it and said, “I am the LORD, the God of Abraham your father and the God of Isaac. The l... | https://www.lesswrong.com/posts/5e7hnJsgpCmvFdGnc/jacob-on-the-precipice |
# Bariatric surgery seems like a no-brainer for most morbidly obese people
Procedures like gastric bypass have been around a long time; long enough that their effects are measured by virtue of their impact on overall health*.* If your goal with weight loss is to lose weight, and not perform a ritual of self-actualizat... | https://www.lesswrong.com/posts/jSTxstnSet3NkTsH8/bariatric-surgery-seems-like-a-no-brainer-for-most-morbidly |
# The Great Disembedding
[](https://substackcdn.com/image/fetch/f_auto,q_auto:good,fl_progressive:steep... | https://www.lesswrong.com/posts/cFaYSpnyjBjoD2nPK/the-great-disembedding |
# The Role of Groups in the Progression of Human Understanding
This post isn't intended to be anything stunningly original, but simply to collect a few basic, but under-utilized observations together in one place.
The question of how we can progress human knowledge and understanding is one that is of particular inter... | https://www.lesswrong.com/posts/xKaJ3YrKL3ms2Bn37/the-role-of-groups-in-the-progression-of-human-understanding |
# What's your standard for good work performance?
A lot of people in the community, including me, are working independently (or have a lot of autonomy, even if they are employed). A lot of people, including me, often feel like they are underperforming or at least wonder if they are. But how do I actually know when I'm... | https://www.lesswrong.com/posts/5xHqtisRZn6PqnapY/what-s-your-standard-for-good-work-performance |
# MMLU’s Moral Scenarios Benchmark Doesn’t Measure What You Think it Measures
In examining the low performance of Large Language Models(LLMs)on the Moral Scenarios task, part of the widely used MMLU benchmark by Hendrycks et al., we found surprising results. When presented with moral scenarios individually, the accura... | https://www.lesswrong.com/posts/XqzWgkP3xekfdh8pa/mmlu-s-moral-scenarios-benchmark-doesn-t-measure-what-you |
# The Hidden Complexity of Wishes - The Animation
This video introduces a series focused on the outer alignment problem. In future videos, we'll explore how this problem affects machine learning systems today and how it could lead to catastrophic outcomes for humanity.
The video is an adaptation of [@Eliezer Yudkowsk... | https://www.lesswrong.com/posts/6At6iMTu3dsKSTXLL/the-hidden-complexity-of-wishes-the-animation |
# Commonsense Good, Creative Good
Let's say you're vegan and you go to a vegan restaurant. The food is quite bad, and you'd normally leave a bad review, but now you're worried: [what if your bad review leads people to go to non-vegan restaurants instead?](https://forum.effectivealtruism.org/posts/v789qKBBQTKNP3mEs/) S... | https://www.lesswrong.com/posts/cEnJsoxQ3ziJEohnP/commonsense-good-creative-good |
# Is Bjorn Lomborg roughly right about climate change policy?
I recently came across Bjorn Lomborg’s 2020 article *Welfare in the 21st century: Increasing development, reducing inequality, the impact of climate change, and the cost of climate policies*. It runs counter to many things I have heard about climate change.... | https://www.lesswrong.com/posts/vWg45pFzaF6a8STJC/is-bjorn-lomborg-roughly-right-about-climate-change-policy |
# Towards Better Milestones for Monitoring AI Capabilities
$$In the present moment, with so much hype around the accomplishments of generative AI, there is a lot of confusion and disagreement regarding timelines for further progress in capabilities. Assumptions regarding human-level AGI range from a few years to at le... | https://www.lesswrong.com/posts/CDJmafoeZJuXHbKDW/towards-better-milestones-for-monitoring-ai-capabilities |
# Projects I would like to see (possibly at AI Safety Camp)
I recently discussed with my AISC co-organiser Remmelt, some possible project ideas I would be excited about seeing at the upcoming AISC, and I thought these would be valuable to share more widely.
*Thanks to Remmelt for helfull suggestions and comments.*
... | https://www.lesswrong.com/posts/J4s5AJ3Xqc8DwAEzQ/projects-i-would-like-to-see-possibly-at-ai-safety-camp |
# Petrov Day Retrospective, 2023 (re: the most important virtue of Petrov Day & unilaterally promoting it)
When Stanislav Petrov's missile alert system pinged, the world was not watching. Russia was not watching. Perhaps a number of superiors in the military were staying in the loop about Stanislav's outpost, waiting ... | https://www.lesswrong.com/posts/c7uQLGrCBLjkbZhPq/petrov-day-retrospective-2023-re-the-most-important-virtue |
# ARC Evals: Responsible Scaling Policies
> We’ve been consulting with several parties[^1^](https://evals.alignment.org/blog/2023-09-26-rsp/#fn:1) on **responsible scaling**[**^2^**](https://evals.alignment.org/blog/2023-09-26-rsp/#fn:2) **policies (RSPs)**. An RSP specifies what level of AI capabilities an AI develop... | https://www.lesswrong.com/posts/pnmFBjHtpfpAc6dPT/arc-evals-responsible-scaling-policies |
# Weighing Animal Worth
It's common for people who approach helping animals from a quantitative direction to need some concept of "[moral weights](https://forum.effectivealtruism.org/topics/moral-weight)" so they can prioritize. If you can avert one year of suffering for a chicken or ten for shrimp which should you ch... | https://www.lesswrong.com/posts/GnDqxxEn3GT2HjdK7/weighing-animal-worth |
# Wobbly Table Theorem in Practice
It turns out the “wobbly table theorem” is less of just a math curiosity than I thought.
The “wobbly table theorem” essentially says that if you have a “table”[^beubn87ppd] resting on a continuous surface, you can find a rotation where the table rests on all four legs[^cwh10ww62pj].... | https://www.lesswrong.com/posts/zDATqYJseBoM9aX5f/wobbly-table-theorem-in-practice |
# Cohabitive Games so Far
A cohabitive game[^dkfa4q3yxns] is a partially cooperative, partially competitive multiplayer game that provides an anarchic dojo for development in applied cooperative bargaining, or negotiation.
Applied cooperative bargaining isn't currently taught, despite being an *infrastructural litera... | https://www.lesswrong.com/posts/bF353RHmuzFQcsokF/cohabitive-games-so-far |
# The point of a game is not to win, and you shouldn't even pretend that it is
*This post started out as a supporting point for* [*another post about games*](https://www.lesswrong.com/posts/bF353RHmuzFQcsokF/peacewagers-so-far)*. It developed into a bit of a practical philosophy post.*
Why do boardgames generally end... | https://www.lesswrong.com/posts/RNbcJedHf4jmZ9GHq/the-point-of-a-game-is-not-to-win-and-you-shouldn-t-even |
# AI #31: It Can Do What Now?
It slices. It dices. Or, at least, it sees, hears, talks, creates stunningly good images and browses the web. Welcome to the newly updated GPT-4. That’s all in two weeks. Throw in Microsoft 365 Copilot finally coming online soon.
Are we back? [I’m guessing we’re back](https://www.youtube... | https://www.lesswrong.com/posts/CeHqm3CSApEjgFb8X/ai-31-it-can-do-what-now |
# Different views of alignment have different consequences for imperfect methods
Almost any powerful AI, with almost any goal, will doom humanity. Hence alignment is often seen as a constraint on AI power: we must direct the AI’s optimisation power in a very narrow direction. If the AI is weak, then imperfect methods ... | https://www.lesswrong.com/posts/wDvHP9KzGqPCtw3PS/different-views-of-alignment-have-different-consequences-for |
# My Current Thoughts on the AI Strategic Landscape
I started working at AI Impacts slightly less than a year ago. Before then, I was not following developments in either AI or AI safety. I do not consider myself a rationalist and did not engage with LessWrong before starting this job. While I have mostly been working... | https://www.lesswrong.com/posts/HX8F7uueCq3G3ESM2/my-current-thoughts-on-the-ai-strategic-landscape |
# Alignment Workshop talks
In February 2023, researchers from a number of top industry AI labs (OpenAI, DeepMind, Anthropic) and universities (Cambridge, NYU) co-organized a two-day workshop on the problem of AI alignment, attended by 80 of the world’s leading machine learning researchers. We’re now making recordings ... | https://www.lesswrong.com/posts/dQ8wiAwnD37y6PkTa/alignment-workshop-talks |
# How to Catch an AI Liar: Lie Detection in Black-Box LLMs by Asking Unrelated Questions
This post is a copy of the introduction of this [paper](https://arxiv.org/abs/2309.15840) on lie detection in LLMs. The Twitter Thread is [here](https://twitter.com/OwainEvans_UK/status/1707451418339377361).
Authors: Lorenzo Pacc... | https://www.lesswrong.com/posts/khFC2a4pLPvGtXAGG/how-to-catch-an-ai-liar-lie-detection-in-black-box-llms-by |
# High-level interpretability: detecting an AI's objectives
*Thanks to Monte MacDiarmid (for discussions, feedback, and experiment infrastructure) and to the Shard Theory team for their prior work and exploratory infrastructure. *
*Thanks to Joseph Bloom, John Wentworth, Alexander Gietelink Oldenziel, Johannes Treuit... | https://www.lesswrong.com/posts/tFYGdq9ivjA3rdaS2/high-level-interpretability-detecting-an-ai-s-objectives |
# The Coming Wave
Book review: The Coming Wave: Technology, Power, and the Twenty-first
Century's Greatest Dilemma, by Mustafa Suleyman.
An author with substantial AI expertise has attempted to discuss AI in
terms that the average book reader can understand.
The key message: AI is about to become possibly the ... | https://www.lesswrong.com/posts/R4byySfuyfv3SGFtJ/the-coming-wave |
# Competitive, Cooperative, and Cohabitive
Summary: I think there's a third genre of games, which I dub Cohabitive games, where your victory isn't tied to other players winning or losing.
(I’ve been writing this in bits and pieces for a while, and [Peacewager](https://www.lesswrong.com/posts/bF353RHmuzFQcsokF/peacewa... | https://www.lesswrong.com/posts/9qJYdFkhNQimvmjBF/competitive-cooperative-and-cohabitive |
# EA Vegan Advocacy is not truthseeking, and it’s everyone’s problem
Introduction
============
Effective altruism prides itself on truthseeking. That pride is justified in the sense that EA is better at truthseeking than most members of its reference category, and unjustified in that it is far from meeting its own st... | https://www.lesswrong.com/posts/aW288uWABwTruBmgF/ea-vegan-advocacy-is-not-truthseeking-and-it-s-everyone-s-1 |
# List of how people have become more hard-working
I've recently asked [how people have become more hard-working](https://www.lesswrong.com/posts/SjvRF88aLJdMdv7RH/how-have-you-become-more-hard-working-1). I compiled the answers across the [EA Forum](https://forum.effectivealtruism.org/posts/bCLoXDbnYSQab8T5t/how-have... | https://www.lesswrong.com/posts/yX7Lfjbfyj2CwMGyH/list-of-how-people-have-become-more-hard-working |
# Steering subsystems: capabilities, agency, and alignment
Human brains have steering subsystems. LLMs and most RL agents do not. [Steering systems](https://www.lesswrong.com/posts/LsqvMKnFRBQh4L3Rs/steering-systems), as defined by Max H, are goal-directed AI systems, or optimizers. Here I focus on steering subsystems... | https://www.lesswrong.com/posts/qzu9o3sTytbC4sZkQ/steering-subsystems-capabilities-agency-and-alignment |
# "Diamondoid bacteria" nanobots: deadly threat or dead-end? A nanotech investigation
*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 am not specifically an expert on the topics covered, so I can’t promise perf... | https://www.lesswrong.com/posts/bc8Ssx5ys6zqu3eq9/diamondoid-bacteria-nanobots-deadly-threat-or-dead-end-a |
# Basic Mathematics of Predictive Coding
This is an overview of the classic 1[999 paper by Rajesh Rao and David Ballard](https://www.nature.com/articles/nn0199_79) that introduced predictive coding. I'm going to focus on [explaining the mathematical setup instead of just staying at a conceptual level](https://www.less... | https://www.lesswrong.com/posts/bD4B2MF7nsGAfH9fj/basic-mathematics-of-predictive-coding |
# Announcing FAR Labs, an AI safety coworking space
FAR Labs is a coworking hub in downtown Berkeley for organizations and individuals working on AI safety and related issues. Since opening the space in March 2023, we have grown to host approximately 30 members. Our members are primarily drawn from four anchor organiz... | https://www.lesswrong.com/posts/bTBmRzYkupcDa6q7X/announcing-far-labs-an-ai-safety-coworking-space |
# Bids To Defer On Value Judgements
Consider two claims:
* “broccoli is good for you”
* “broccoli decreases cholesterol”
Even though the former might be considered a lossy summary of the latter, the two feel very different; they pull very different levers in my brain. “Broccoli decreases cholesterol” pulls lever... | https://www.lesswrong.com/posts/QsPfngERTPJ3eXPBx/bids-to-defer-on-value-judgements |
# The Retroactive Funding Landscape: Innovations for Donors and Grantmakers
**Summary:** This is an overview of existing solutions for retroactive funding – also known as outcomes-based funding: What are the main problems that each solution targets and how does it try to solve them? You may learn of new funding models... | https://www.lesswrong.com/posts/NcgrLh4hgEBb9DuTT/the-retroactive-funding-landscape-innovations-for-donors-and |
# Paul Christiano's views on "doom" (video explainer)
I have just published a [video explainer](https://youtu.be/K8SUBNPAJnE) of Paul Christiano's Lesswrong post: [My views on "doom".](https://www.lesswrong.com/posts/xWMqsvHapP3nwdSW8/my-views-on-doom)
This is the second Lesswrong post I make a video explainer about,... | https://www.lesswrong.com/posts/P77rbnsgF5f9kspk4/paul-christiano-s-views-on-doom-video-explainer |
# Headphones hook
Some months ago I got a 3D printer. (An Anycubic Kobra Go, which was pretty high up the list of "best value for money for a starter printer" that I found through [/r/3Dprinting](https://www.reddit.com/r/3Dprinting/) at the time.) I haven't used it much, but recently I used it to solve an itch. I want... | https://www.lesswrong.com/posts/bi7LXfFtpM8cLWSrg/headphones-hook |
# The Lighthaven Campus is open for bookings
Lightcone Infrastructure (the organization that grew from and houses the LessWrong team) has just finished renovating a 7-building physical campus that we hope to use to make the future of humanity go better than it would otherwise.
We're hereby announcing that it is gener... | https://www.lesswrong.com/posts/memqyjNCpeDrveayx/the-lighthaven-campus-is-open-for-bookings |
# My submission to the ALTER Prize
This is a linkpost for my submission to the ALTER AI Alignment Prize. As such, it's 8 and a half pages of pretty unapologetically dense math and/or philosophy the whole way through. It represents my own work, and I stand by the results and thinking in it. As such, I welcome actionabl... | https://www.lesswrong.com/posts/u6xsHmhfaQ8YA8fjw/my-submission-to-the-alter-prize |
# How model editing could help with the alignment problem
### Preface
This article explores the potential of model editing techniques in aligning future AI systems. Initially, I was skeptical about its efficacy, especially considering the objectives of current model editing methods. I argue that merely editing "fact... | https://www.lesswrong.com/posts/FNmzn33akiSesc4Ke/how-model-editing-could-help-with-the-alignment-problem |
# Conditionals All The Way Down
*(I thought about this idea on my own before Googling to see if anyone had already written it up. I found something very similar at* [*https://projecteuclid.org/journals/notre-dame-journal-of-formal-logic/volume-50/issue-2/Justification-by-an-Infinity-of-Conditional-Probabilities/10.121... | https://www.lesswrong.com/posts/Ys6hZBvKGBD2FnPqj/conditionals-all-the-way-down |
# Arguments for moral indefinability
*Epistemic status: I endorse the core intuitions behind this post, but am only moderately confident in the specific claims made. Also, while I do have a degree in philosophy, I am not a professional ethicist, and I’d appreciate feedback on how these ideas relate to existing literat... | https://www.lesswrong.com/posts/fmkPPpmLKKJRK4AR2/arguments-for-moral-indefinability-1 |
# My Effortless Weightloss Story: A Quick Runthrough
This is Part I in a [series on easy weightloss without any need for will power. ](https://www.lesswrong.com/s/wbPTdgCragwkFNc2T)
The Origin: listening to the dark corners of the internet
---------------------------------------------------------
Losing weight is su... | https://www.lesswrong.com/posts/tkTwFAmrCx45Yn8hY/my-effortless-weightloss-story-a-quick-runthrough |
# New Tool: the Residual Stream Viewer
This is a link-post for the residual stream viewer, which can be found [here](https://tinyurl.com/resid-viewer). It's an online tool whose goal is to make it easier to do interpretability research by letting you easily look at directions within the residual stream. It's still in ... | https://www.lesswrong.com/posts/yawkDNfjR2nnzycMZ/new-tool-the-residual-stream-viewer |
# "Absence of Evidence is Not Evidence of Absence" As a Limit
Here are two cases involving absence of evidence:
* You wait at the bus stop for a half hour without seeing a bus, even though this bus is supposed to come every 10 minutes. This is strong evidence against the hypothesis that the buses are running today (p... | https://www.lesswrong.com/posts/YcdZmT7txibjK5jPG/absence-of-evidence-is-not-evidence-of-absence-as-a-limit |
# AI Safety Impact Markets: Your Charity Evaluator for AI Safety
**Summary:** This is a quick overview of the current state of our system *AI Safety Impact Markets*: We want to give donors an improved ability to coordinate so that their diverse knowledge can inform our crowdsourced donation recommendations. AI Safety ... | https://www.lesswrong.com/posts/DygTMPmFeTG7mFCB7/ai-safety-impact-markets-your-charity-evaluator-for-ai |
# Fifty Flips
*An unfair coin (potentially EXTREMELY unfair) will be flipped fifty times. Your goal is to correctly predict as many of these flips as possible, by deducing the nature of the unfairness as quickly as possible.*
*\[Predict Heads\] \[Predict Tails\]*
* * *
You can play this (in-browser, very short) gam... | https://www.lesswrong.com/posts/hsA8hnvuyaXZdx3am/fifty-flips |
# Join AISafety.info's Distillation Hackathon (Oct 6-9th)
tl;dr: Contribute to[ aisafety.info](https://aisafety.info) by answering questions about AI Safety from October 6th to October 9th. Participation in hackathons is the basis for applying to future fellowships, and there are prizes to be won by the top entrants. ... | https://www.lesswrong.com/posts/Re8ECqwpzcpCAkrMo/join-aisafety-info-s-distillation-hackathon-oct-6-9th |
# Revisiting the Manifold Hypothesis
*In the following analysis, a machine learning challenge is proposed by Sasha Kolpakov, Managing Editor of the Journal of Experimental Mathematics, and myself to verify a strong form of the Manifold Hypothesis due to Yoshua Bengio. *
Motivation:
------------
In the Deep Learning... | https://www.lesswrong.com/posts/uH4GfZwBYteiy2GrL/revisiting-the-manifold-hypothesis |
# Why I got the smallpox vaccine in 2023
A common figure cited for the death toll of smallpox is 500 million, as in the title of the [excellent essay](https://forum.effectivealtruism.org/posts/jk7A3NMdbxp65kcJJ/500-million-but-not-a-single-one-more) and [adaptation](https://www.youtube.com/watch?v=ljmifo4Klss). Half a... | https://www.lesswrong.com/posts/QZHL9kmuBfioRAQYM/why-i-got-the-smallpox-vaccine-in-2023 |
# Linkpost: They Studied Dishonesty. Was Their Work a Lie?
This is a linkpost for [**Gideon Lewis-Kraus**](https://www.newyorker.com/contributors/gideon-lewis-kraus)**'s** New Yorker article on the (alleged) Ariely and Gino data fraud scandals. I've been following this situation off-and-on for a while (and even more ... | https://www.lesswrong.com/posts/LizpBcsF9XEAhTsvn/linkpost-they-studied-dishonesty-was-their-work-a-lie |
# The 99% principle for personal problems
Often when people are dealing with an issue – emotional, mental, or physical – there’s genuine progress and the issue becomes less and seems to go away.
Until it comes back, seemingly as bad as before.
Maybe the person developed a coping mechanism that worked, but only under... | https://www.lesswrong.com/posts/7jufC5pdJx8CrHv3J/the-99-principle-for-personal-problems |
# Direction of Fit
This concept has recently become a core part of my toolkit for thinking about the world, and I find it helps explain a lot of things that previously felt confusing to me. Here I explain how I understand “direction of fit,” and give some examples of where I find the concept can be useful.
Handshake... | https://www.lesswrong.com/posts/6xKMSfK8oTpTtWKZN/direction-of-fit-1 |
# energy landscapes of experts
Suppose you're a choosing an expert for an important project. One approach is to choose a professor at a prestigious university whose research is superficially related to the project, and ask them to recommend someone. People have a better understanding of some conceptual and social area... | https://www.lesswrong.com/posts/LBo3ZJhtX8eJz75HK/energy-landscapes-of-experts |
# Will early transformative AIs primarily use text? [Manifold question]
Will early transformative AIs primarily use text to store intermediate thoughts and memories?
I think the answer to this question matters for alignment research prioritization, and getting some common knowledge about what are the most commonly he... | https://www.lesswrong.com/posts/rcsKqFAqrsZSkRCtJ/will-early-transformative-ais-primarily-use-text-manifold |
# A counterexample for measurable factor spaces
[PDF](https://drive.google.com/file/d/1h4zVKuPZggdn_dJcDo0v_sTH_7o9AwAe/view?usp=sharing)
This is a technical post researching infinite [factor spaces](https://www.lesswrong.com/posts/nksgLdLc7y5KtbmFw/a-reformulation-of-finite-factored-sets)
I define a general measura... | https://www.lesswrong.com/posts/4WERwizaeiEnFbNHp/a-counterexample-for-measurable-factor-spaces-1 |
# Expectations for Gemini: hopefully not a big deal
**Introduction**
----------------
My goal is to register and share my expectations and hear others' opinions on their expectation for the relative performances of Gemini VS GPT-4.
**My expectations**
-------------------
GPT-4 to Gemini will likely not be as big a ... | https://www.lesswrong.com/posts/PxELfZnvbv8jcKewp/expectations-for-gemini-hopefully-not-a-big-deal |
# Dall-E 3
It seems like we do not yet have a post about Dall-E 3. It was announced a few days ago and can now be tried via bing image creator (long wait times).
I think it is worth mentioning, because it seems like a big jump.
I spend quite a lot of time playing around with Dalle mini, Dall-E 2, Stable Diffusion ... | https://www.lesswrong.com/posts/FaK8irvBhdunE33tF/dall-e-3 |
# My Mid-Career Transition into Biosecurity
After working as a professional programmer for fourteen years, primarily in ads and web performance, I [switched careers](https://www.jefftk.com/p/leaving-google-joining-the-nucleic-acid-observatory) to biosecurity. It's now been a bit over a year: how has it gone?
In terms... | https://www.lesswrong.com/posts/6NDKuaKo6MctWyJ4o/my-mid-career-transition-into-biosecurity |
# Some Quick Follow-Up Experiments to “Taken out of context: On measuring situational awareness in LLMs”
Introduction
============
A team working with Owain Evans recently released [Taken out of context: On measuring situational awareness in LLMs](https://arxiv.org/abs/2309.00667). I think this work is very exciting.... | https://www.lesswrong.com/posts/YRis8ZDstqnaW2erL/some-quick-follow-up-experiments-to-taken-out-of-context-on |
# Early Experiments in Reward Model Interpretation Using Sparse Autoencoders
*This research was performed by* [*Luke Marks*](https://www.lesswrong.com/users/marc-er)*,* [*Amirali Abdullah*](https://www.lesswrong.com/users/amirali-abdullah)*,* [*nothoughtsheadempty*](https://www.lesswrong.com/users/nothoughtsheadempty)... | https://www.lesswrong.com/posts/QXEeis95sKrStLu2Q/early-experiments-in-reward-model-interpretation-using |
# Mech Interp Challenge: October - Deciphering the Sorted List Model
**I'm writing this post to discuss solutions to the September challenge, and present the challenge for this October.**
If you've not read the [first post in this sequence](https://www.alignmentforum.org/posts/dpRuey9MDHNACSDtn/mech-interp-challenge... | https://www.lesswrong.com/posts/frLTfKr8NFv7WCcWG/mech-interp-challenge-october-deciphering-the-sorted-list |
# Monthly Roundup #11: October 2023
It never stops. I’m increasingly building distinct roundups for various topics, in particular I’m splitting medical and health news out. Let’s get to the rest of it.
### Bad News
[A simple model of why everything sucks](https://nothinghuman.substack.com/p/the-tyranny-of-the-margin... | https://www.lesswrong.com/posts/LszTjZCb4toYrfiAh/monthly-roundup-11-october-2023 |
# Potential alignment targets for a sovereign superintelligent AI
I'd like to compile a list of potential alignment targets for a [sovereign](https://www.lesswrong.com/posts/yTy2Fp8Wm7m8rHHz5/superintelligence-15-oracles-genies-and-sovereigns) superintelligent AI.
By an alignment target, I mean something like what go... | https://www.lesswrong.com/posts/bcetbJdD6eJ9terYL/potential-alignment-targets-for-a-sovereign-superintelligent |
# What would it mean to understand how a large language model (LLM) works? Some quick notes.
*Cross-posted from* [*New Savanna*](https://new-savanna.blogspot.com/2023/10/what-would-it-mean-to-understand-how.html)*.*
I don’t mean “understand” in any deep philosophical sense. I mean only a rough and ready sense of the... | https://www.lesswrong.com/posts/Be3CfAW5PMWT9nNY9/what-would-it-mean-to-understand-how-a-large-language-model |
# Metaculus Announces Forecasting Tournament to Evaluate Focused Research Organizations, in Partnership With the Federation of American Scientists
You are invited to participate in Metaculus's [FRO-Casting Tournament](https://www.metaculus.com/tournament/fro-casting/), an exciting pilot project in partnership with [T... | https://www.lesswrong.com/posts/RNEs9iEn8xoa4Rpga/metaculus-announces-forecasting-tournament-to-evaluate |
# Current AI safety techniques?
What safety techniques exist? What techniques should labs be using, now or if they soon develop powerful models? What techniques should be considered industry-best-practices? "How might we align transformative AI if it’s developed very soon?" What's our current bag of tricks? What align... | https://www.lesswrong.com/posts/qMWLkLfuxgeWzB26F/current-ai-safety-techniques |
# OpenAI-Microsoft partnership
OpenAI has a [strong](https://openai.com/blog/openai-and-microsoft-extend-partnership) [partnership](https://blogs.microsoft.com/blog/2023/01/23/microsoftandopenaiextendpartnership/) with Microsoft. The details are opaque, as far as I know. It tentatively seems that OpenAI is required to... | https://www.lesswrong.com/posts/yvEzpn6vpMGDLxXC9/openai-microsoft-partnership |
# When to Get the Booster?
Let's say you're planning on getting a covid booster this fall: when's the best time to get it? The updated boosters ( [targeting the XBB lineage](https://www.cdc.gov/coronavirus/2019-ncov/vaccines/stay-up-to-date.html#updated-vaccines)) have been out since mid September, so there's not a ne... | https://www.lesswrong.com/posts/N4tCNY6uEv4R99KkN/when-to-get-the-booster |
# AXRP Episode 25 - Cooperative AI with Caspar Oesterheld
[YouTube link](https://youtu.be/0JkaOAzDfgE)
Imagine a world where there are many powerful AI systems, working at cross purposes. You could suppose that different governments use AIs to manage their militaries, or simply that many powerful AIs have their own w... | https://www.lesswrong.com/posts/CTh3JHBmEfHjE7WP5/axrp-episode-25-cooperative-ai-with-caspar-oesterheld |
# [Link] Bay Area Winter Solstice 2023
Bay Area Winter Solstice 2023 will be held on December 16 at the Freight & Salvage in downtown Berkeley!
Tickets are available at [https://www.eventbrite.com/e/bay-area-winter-solstice-2023-tickets-721678057497.](https://www.eventbrite.com/e/bay-area-winter-solstice-2023-tickets... | https://www.lesswrong.com/posts/KKvBDhbpMj7cPpL2L/link-bay-area-winter-solstice-2023 |
# Graphical tensor notation for interpretability
\[ This post is now on arXiv too: [https://arxiv.org/abs/2402.01790](https://arxiv.org/abs/2402.01790) \]
Some examples of graphical tensor notation from... | https://www.lesswrong.com/posts/BQKKQiBmc63fwjDrj/graphical-tensor-notation-for-interpretability |
# What evidence is there of LLM's containing world models?
Since this still seems to be an area of active debate within the ML community, it may be worthwhile providing a location to gather this evidence all in one place. Please only list one paper per answer as that makes it easier for people to comment on it (and po... | https://www.lesswrong.com/posts/2MpX7eR6qLTDmyeMJ/what-evidence-is-there-of-llm-s-containing-world-models |
# I don’t find the lie detection results that surprising (by an author of the paper)
(This is just my opinion, not necessarily shared by the other co-authors)
Many people have commented that they find the results of our recent [lie detection paper](https://www.lesswrong.com/posts/khFC2a4pLPvGtXAGG/how-to-catch-an-ai-... | https://www.lesswrong.com/posts/noxJrzXcdz738uqMi/i-don-t-find-the-lie-detection-results-that-surprising-by-an |
# AISN #23:
New OpenAI Models, News from Anthropic, and Representation Engineering
Welcome to the AI Safety Newsletter by the [Center for AI Safety](https://www.safe.ai/). We discuss developments in AI and AI safety. No technical background required.
Subscribe [here](https://newsletter.safe.ai/subscribe?utm_medium=we... | https://www.lesswrong.com/posts/oAu6nJgAzN5LJZsmg/aisn-23-new-openai-models-news-from-anthropic-and |
# How to solve deception and still fail.
*A mostly finished post I'm kicking out the door. You'll get the gist.*
**I**
There's a tempting picture of alignment that centers on the feeling of "As long as humans stay in control, it will be okay." Humans staying in control, in this picture, is something like humans givi... | https://www.lesswrong.com/posts/dWMzzd6hfimTQk8yk/how-to-solve-deception-and-still-fail |
# Fraternal Birth Order Effect and the Maternal Immune Hypothesis
TLDR; The Bogaert et all (2017) study which claims to find evidence that a particular protein and corresponding antibodies contribute to the FBOE has a significant problem which contradicts the main claim.
(I had investigated the below a few years ago ... | https://www.lesswrong.com/posts/ngQke8zyPBkp4J4tt/fraternal-birth-order-effect-and-the-maternal-immune |
# How to Get Rationalist Feedback
For the vague ideas and posts that you *want* to write up, but which don't feel "ready to be A Real LessWrong Post" yet.
- Describe your idea on a LessWrong or LessWrong-adjacent Discord server. The adjacent servers (in my experience) are more active. For AI safety/alignment ideas sp... | https://www.lesswrong.com/posts/NwWhbgovCwGQCKAZi/how-to-get-rationalist-feedback |
# Response to Quintin Pope's Evolution Provides No Evidence For the Sharp Left Turn
Response to: [Evolution Provides No Evidence For the Sharp Left Turn](https://www.lesswrong.com/posts/hvz9qjWyv8cLX9JJR/evolution-provides-no-evidence-for-the-sharp-left-turn), due to it winning first prize in [The Open Philanthropy Wo... | https://www.lesswrong.com/posts/Wr7N9ji36EvvvrqJK/response-to-quintin-pope-s-evolution-provides-no-evidence |
# Can the House Legislate?
The US government is in a less functional position than usual:
* Legislation requires (among other things) a plurality vote of the House of Representatives.
* This requires the Speaker of the House to call a vote.
* This requires a Speaker of the House.
* The House re... | https://www.lesswrong.com/posts/3hSev6ZrnjdnkkiZv/can-the-house-legislate |
# AI #32: Lie Detector
Can you tell if an AI is lying to you? A new paper claims that we essentially can do exactly that, at least under the right conditions. Another paper claims we can inject various sentiments into responses, getting the AI to do what we wish. Interpretability is making progress. It is exciting to ... | https://www.lesswrong.com/posts/iDoTRmCH22PaFYL6x/ai-32-lie-detector |
# This anime storyboard doesn't exist: a graphic novel written and illustrated by GPT4
Six months ago, GPT4 wrote the first known AI-written novella, "[*The Troodon Quill*](https://www.lesswrong.com/posts/7AxzaEDP8WWjEouSA/gpt4-is-capable-of-writing-decent-long-form-science-fiction-1)".
With the recent release of DA... | https://www.lesswrong.com/posts/pQzRj4hJRtMxg3hib/this-anime-storyboard-doesn-t-exist-a-graphic-novel-written |
# Translations Should Invert
Suppose that Alice is arguing for ZFC set theory as the proper foundation of mathematics. Bob is arguing for some specific intuitionistic type theory. Alice uses classical logic, and Bob uses intuitionistic logic; this is a big part of their difference in thinking, but not the whole disagr... | https://www.lesswrong.com/posts/PNiggFr7gDRyavtRH/translations-should-invert |
# Evaluating the historical value misspecification argument
ETA: I'm not saying that MIRI thought AIs wouldn't understand human values. If there's only one thing you take away from this post, please don't take away that. [Here is Linch's attempted summary of this post](https://www.lesswrong.com/posts/i5kijcjFJD6bn7dwq... | https://www.lesswrong.com/posts/i5kijcjFJD6bn7dwq/evaluating-the-historical-value-misspecification-argument |
# A well-defined history in measurable factor spaces
[PDF](https://drive.google.com/file/d/1iLtE3lxQQcW6ltIn_umXw1hiI5EqvnoV/view?usp=sharing)
This is a technical post researching infinite [factor spaces](https://www.lesswrong.com/posts/nksgLdLc7y5KtbmFw/a-reformulation-of-finite-factored-sets)
In a factor space $F ... | https://www.lesswrong.com/posts/EKPSgN8LsiEJzX5ni/a-well-defined-history-in-measurable-factor-spaces |
# Ideation and Trajectory Modelling in Language Models
*\[Epistemic Status: Exploratory, and I may have confusions\]*
**Introduction**
================
LLMs and other possible RL agent have the property of taking many actions iteratively. However, not all possible short-term outputs are equally likely, and I think b... | https://www.lesswrong.com/posts/j9qG76qAKygPbGqZy/ideation-and-trajectory-modelling-in-language-models |
# Towards Monosemanticity: Decomposing Language Models With Dictionary Learning
*Text of post based on [our blog post](https://www.anthropic.com/index/decomposing-language-models-into-understandable-components) as [a linkpost for the full paper](https://transformer-circuits.pub/2023/monosemantic-features/) which is co... | https://www.lesswrong.com/posts/TDqvQFks6TWutJEKu/towards-monosemanticity-decomposing-language-models-with |
# Stampy's AI Safety Info soft launch
[Stampy’s AI Safety Info](https://aisafety.info) is a project to create an interactive FAQ about existential risk from AI, started by [Rob Miles](https://www.youtube.com/c/robertmilesai). Our goal is to build a single resource aimed at informing all audiences, whether that means g... | https://www.lesswrong.com/posts/obMiQv9K76nRZj9tE/stampy-s-ai-safety-info-soft-launch |
# Provably Safe AI
I've been hearing vague claims that automated theorem provers are able
to, or will soon be able to prove things about complex software such as
AIs.
Max Tegmark and Steve Omohundro have now published a paper, [Provably
Safe Systems : The Only Path To Controllable
AGI](https://arxiv.org/abs/230... | https://www.lesswrong.com/posts/KX3Qwr7QM7CvhJLG6/provably-safe-ai |
# The Long-Term Future Fund is looking for a full-time fund chair
**EDIT 2023/11/21**
**We have finished evaluating the first batch of our applications.**
**As we have not yet finished the hiring process, people are still welcome to apply to us. We are still very likely to skim future applications, however we have n... | https://www.lesswrong.com/posts/Zu8K3iyAPcsMRvXZb/the-long-term-future-fund-is-looking-for-a-full-time-fund |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.