text stringlengths 300 320k | source stringlengths 52 154 |
|---|---|
# Preface to the Sequence on LLM Psychology
As Large Language Models (LLMs) like ChatGPT evolve, becoming more advanced and intricate, the challenge of understanding their behaviors becomes increasingly hard. Solely relying on traditional interpretability techniques [may not be sufficient](https://www.lesswrong.com/po... | https://www.lesswrong.com/posts/yuwdj82yjhLFYessc/preface-to-the-sequence-on-llm-psychology |
# The Stochastic Parrot Hypothesis is debatable for the last generation of LLMs
*This post is part of a sequence on LLM Psychology.*
[*@Pierre Peigné*](https://www.lesswrong.com/users/pierre-peigne?mention=user) *wrote the details section in argument 3 and the other weird phenomenon. The rest is written in the voice ... | https://www.lesswrong.com/posts/HxRjHq3QG8vcYy4yy/the-stochastic-parrot-hypothesis-is-debatable-for-the-last |
# AMA: Earning to Give
This week the [Effective Altruism Forum](https://forum.effectivealtruism.org/) is running an [Effective Giving Spotlight](https://forum.effectivealtruism.org/posts/hAzhyikPnLnMXweXG/participate-in-the-donation-election-and-the-first-weekly#First_weekly_theme__Effective_Giving_Spotlight__November... | https://www.lesswrong.com/posts/nkJDNJB5g2S7ssiwP/ama-earning-to-give |
# Mirror, Mirror on the Wall: How Do Forecasters Fare by Their Own Call?
*Note: This is a* [*linkpost*](https://www.metaculus.com/notebooks/19812/mirror-mirror-on-the-wall-how-do-forecasters-fare-by-their-own-call/) *for the Metaculus Journal. I work for Metaculus and this is part of a* [*broader exploration*](https:/... | https://www.lesswrong.com/posts/B7yefKWDHKMhDanP4/mirror-mirror-on-the-wall-how-do-forecasters-fare-by-their |
# Implementing Decision Theory
I'm _implementing_ decision theory, in Python. There are three parts to this:
1. An XML language for formally specifying decision theory dilemmas. It can express many dilemmas: Newcomb, Transparent Newcomb, Parfit's Hitchhiker, Death in Damascus, Smoking Lesion, Cosmic Ray, symmetric Pr... | https://www.lesswrong.com/posts/bqbL4pt92AGCCcvQH/implementing-decision-theory |
# Scalable And Transferable Black-Box Jailbreaks For Language Models Via Persona Modulation
**Paper coauthors:** [Rusheb Shah](https://www.linkedin.com/in/ACoAAB63nTsBLsxbiGUom3-xG_DLLJvBLPnwn0M), [Quentin Feuillade--Montixi](https://www.linkedin.com/in/ACoAADL_UysBW9lkWi6Q_x8qEjF4k2fk9F2qlls), [Soroush J. Pour](https... | https://www.lesswrong.com/posts/kaxqjCKJL6RNHwJLD/scalable-and-transferable-black-box-jailbreaks-for-language-2 |
# Vote on Interesting Disagreements
Do you have a question you'd like to see argued about? Would you like to indicate your position and [discuss](https://www.lesswrong.com/dialogues) it with someone who disagrees?
Add poll options to [the thread below](https://www.lesswrong.com/posts/hc9nMipTXy2sm3tJb/disagreements-... | https://www.lesswrong.com/posts/hc9nMipTXy2sm3tJb/vote-on-interesting-disagreements |
# Announcing Athena - Women in AI Alignment Research
Athena is a new research mentorship program fostering diversity of ideas in AI safety research. We aim to get more women and marginalized genders into technical research and offer the support needed to thrive in this space.
Applications for scholars are open until... | https://www.lesswrong.com/posts/pJ9qWeBRRuvPvnoNK/announcing-athena-women-in-ai-alignment-research |
# Progress links digest, 2023-11-07: Techno-optimism and more
Again I’ve been delayed in putting this out by travel etc., so it’s a longer one with links from the last month or so:
**On the Progress Forum**
-------------------------
* [Radical Energy Abundance](https://progressforum.org/posts/y4kYmFhqmA6gxsu9Y/rad... | https://www.lesswrong.com/posts/qgfBuaKe8fhWjkGvb/progress-links-digest-2023-11-07-techno-optimism-and-more |
# South Bay ACX/LW/EA Meetup & Vegansgiving Potluck
Ilmarin House is hosting an ACX/LW/EA meetup and plant-based potluck. These are sort of parallel events; the meetup will take place as usual (ie unstructured conversation) except those who want to are invited to join us at the dinner table at five.
If you can't bri... | https://www.lesswrong.com/events/Df5mLGpnyyQkH7mwR/south-bay-acx-lw-ea-meetup-and-vegansgiving-potluck |
# Concrete positive visions for a future without AGI
> *"There was a threshold crossed somewhere," said the Confessor, "without a single apocalypse to mark it. Fewer wars. Less starvation. Better technology. The economy kept growing. People had more resource to spare for charity, and the altruists had fewer and f... | https://www.lesswrong.com/posts/rsTwJGdPqipt7PJs3/concrete-positive-visions-for-a-future-without-agi |
# Quantopian contest, but for food intake and weight
Twelve years ago, I lost 100 lbs. [in a fairly boring manner](https://www.reddit.com/r/loseit/comments/498afc/) by eating 1200 calories a day. Brutally unpleasant and catabolic, but mostly successful. It's creeping back. There are no low-hanging fruits like sugarwat... | https://www.lesswrong.com/posts/v2LazrqH5odiYucaG/quantopian-contest-but-for-food-intake-and-weight |
# Impressions from base-GPT-4?
I wonder if some people here had a chance to play with base-GPT-4 (the access is given very selectively for research purposes) and would not mind sharing some of their impressions?
I know that some people have been playing with it, but I've never seen a discussion of impressions and le... | https://www.lesswrong.com/posts/tbJdxJMAiehewGpq2/impressions-from-base-gpt-4 |
# Thinking By The Clock
I.
--
I'm sure Harry Potter and the Methods of Rationality taught me some of the obvious, overt things it set out to teach. Looking back on it a decade after I first read it however, what strikes me most strongly are often the brief, tossed off bits in the middle of the flow of a story.
> Fre... | https://www.lesswrong.com/posts/WJtq4DoyT9ovPyHjH/thinking-by-the-clock |
# Running your own workshop on handling hostile disagreements
*tl;dr: this is a guide for running a productive disagreement workshop. Decide of modalities, choose a few techniques, study, practice, make some post-workshop materials and be cautious.*
*Epistemic Status: Quickly written. Literally reporting what I start... | https://www.lesswrong.com/posts/xQgsdEn6vwWynx4Gi/running-your-own-workshop-on-handling-hostile-disagreements |
# Growth and Form in a Toy Model of Superposition
> ***TLDR:** This post distills* [Dynamical and Bayesian Phase Transitions in a Toy Model of Superposition](https://arxiv.org/abs/2310.06301) *by Chen et al. (2023), where they study developmental stages of the Toy Model of Superposition, understanding growth and form ... | https://www.lesswrong.com/posts/jvGqQGDrYzZM4MyaN/growth-and-form-in-a-toy-model-of-superposition |
# How well does your research adress the theory-practice gap?
(5-6 min read time without exercise instructions)
***Motivation for post**: Finding a way to map mathematics & areas of academic research to the alignment problem seems to be a core part of theoretical alignment and agent foundations work. Knowing how to f... | https://www.lesswrong.com/posts/kGopQHxeKmJm3iXSe/how-well-does-your-research-adress-the-theory-practice-gap |
# Tall Tales at Different Scales: Evaluating Scaling Trends For Deception In Language Models
*This post summarizes work done over the summer as part of the* [*Summer 2023 AI Safety Hub Labs programme*](https://www.aisafetyhub.org/)*. Our results will also be published as part of an upcoming paper. In this post, we foc... | https://www.lesswrong.com/posts/pip63HtEAxHGfSEGk/tall-tales-at-different-scales-evaluating-scaling-trends-for |
# Can a stupid person become intelligent?
I've come to a rather uncomfortable self-assessment: I believe I am a stupid person. This isn't an easy thing to say, especially in a community like LessWrong, where intellect and deep thinking are highly valued. But it's a sentiment that has been echoing in my head for a whil... | https://www.lesswrong.com/posts/AdKzf7r8RGj4vnWYE/can-a-stupid-person-become-intelligent |
# Five projects from AI Safety Hub Labs 2023
AI Safety Hub Labs is a research programme that helps early-career researchers to complete an AI safety research project. Projects are completed in groups of 3-5 participants, supervised by a more senior safety researcher, and managed by AI Safety Hub. This summer’s program... | https://www.lesswrong.com/posts/rT6uHEN7ddZAmwbJv/five-projects-from-ai-safety-hub-labs-2023-1 |
# Is there a lesswrong archive of all public posts?
like for example: the [wikimedia db dumps](https://dumps.wikimedia.org/), or the [stack exchange db dumps](https://archive.org/details/stackexchange).
i'd like to be able to browse lesswrong while offline. even better if i can process the data with a script while of... | https://www.lesswrong.com/posts/aoT4TTScmpPXxCWas/is-there-a-lesswrong-archive-of-all-public-posts |
# Why is lesswrong blocking wget and curl (scrape)?
if there is no official lesswrong db/site archive for public posts, i'd like to be able to create my own with automated tools like [wget](https://en.wikipedia.org/wiki/Wget), so that i can browse the site while offline.
see [Is there a lesswrong archive of all public... | https://www.lesswrong.com/posts/gidrFxE5hdQWCrXxn/why-is-lesswrong-blocking-wget-and-curl-scrape |
# Open Agency model can solve the AI regulation dilemma
It seems to me that most people concerned about AI regulation (and calls for "CERN for AI", or the proposals such as the [OAA](https://www.lesswrong.com/posts/jRf4WENQnhssCb6mJ/davidad-s-bold-plan-for-alignment-an-in-depth-explanation)) are concerned about the mo... | https://www.lesswrong.com/posts/CqYaazaG6EkovspMT/open-agency-model-can-solve-the-ai-regulation-dilemma |
# Metaculus's New Sidebar Helps You Find Forecasts Faster
Metaculus features over 8,000 forecast questions. We’ve introduced a new sidebar to make it easy to find and focus on the ones you care about most.
 was this week. What delicious and/or terrifying things await?
#### Turbo Boost
First off, we have GPT-4-Turbo.
> Today we’re launching a preview of the next generation of this model, [GPT-4 Turbo](http... | https://www.lesswrong.com/posts/wdekcGpsMtakGCo5y/on-openai-dev-day |
# Polysemantic Attention Head in a 4-Layer Transformer
*Produced as a part of MATS Program, under* [*@Neel Nanda*](https://www.lesswrong.com/users/neel-nanda-1?mention=user) *and* [*@Lee Sharkey*](https://www.lesswrong.com/users/lee_sharkey?mention=user) *mentorship*
***Epistemic status:** optimized to get the post o... | https://www.lesswrong.com/posts/nuJFTS5iiJKT5G5yh/polysemantic-attention-head-in-a-4-layer-transformer |
# Learning-theoretic agenda reading list
Recently, I'm receiving more and more requests for a self-study reading list for people interested in the [learning-theoretic agenda](https://www.alignmentforum.org/posts/ZwshvqiqCvXPsZEct/the-learning-theoretic-agenda-status-2023). I created a standard list for that, but befor... | https://www.lesswrong.com/posts/fsGEyCYhqs7AWwdCe/learning-theoretic-agenda-reading-list |
# AI #37: Moving Too Fast
We [had OpenAI’s dev day](https://thezvi.substack.com/p/on-openai-dev-day), where they introduced a host of new incremental feature upgrades including a longer context window, more recent knowledge cutoff, increased speed, seamless feature integration and a price drop. Quite the package. On t... | https://www.lesswrong.com/posts/44Cv4HFoWEZvFnL5u/ai-37-moving-too-fast |
# Text Posts from the Kids Group: 2021
Another round of liberating kid posts from [Facebook](https://www.jefftk.com/p/making-groups-for-kid-pictures). For reference, in 2021 Lily turned 7, Anna turned 5, and Nora was born.
(Some of these were from me; some were from Julia. Ones saying "me" could mean either of us.)
... | https://www.lesswrong.com/posts/xmDzYvasaegWvFWtx/text-posts-from-the-kids-group-2021 |
# How I got so excited about HowTruthful
*This is the script for a* [*video*](https://www.youtube.com/watch?v=XqoJAyihJ_c) *I made about my current full-time project. I think the LW community will understand its value better than the average person I talk to does.*
Hi, I'm Bruce Lewis. I'm a computer programmer. For ... | https://www.lesswrong.com/posts/f2CftRRndH97RpAwL/how-i-got-so-excited-about-howtruthful |
# ACI#6: A Non-Dualistic ACI Model
Most traditional AI models are dualistic. As [Demski & Garrabrant have pointed out](https://www.lesswrong.com/s/Rm6oQRJJmhGCcLvxh/p/i3BTagvt3HbPMx6PN), these models assume that an agent is an object that persists over time, and has well-defined input/output channels, like it's playin... | https://www.lesswrong.com/posts/FRd6nNj3M33w2CSX5/aci-6-a-non-dualistic-aci-model |
# AI Timelines
Introduction
============
How many years will pass before transformative AI is built? Three people who have thought about this question a lot are Ajeya Cotra from [Open Philanthropy](https://www.openphilanthropy.org/), Daniel Kokotajlo from [OpenAI](https://openai.com/) and Ege Erdil from [Epoch](https... | https://www.lesswrong.com/posts/K2D45BNxnZjdpSX2j/ai-timelines |
# Crock, Crocker, Crockiest
**Summary**: If you declare you’re operating by Crocker’s Rules, other people are allowed to optimize their messages to you for information, not for being nice to you. The converse of Crocker’s Rules would be asking people to optimize their messages to you for being nice to you, not for inf... | https://www.lesswrong.com/posts/BcCeyL89cKSqcjtL5/crock-crocker-crockiest |
# GPT-2030 and Catastrophic Drives: Four Vignettes
I previously [discussed the capabilities](https://bounded-regret.ghost.io/what-will-gpt-2030-look-like/) we might expect from future AI systems, illustrated through GPT~2030~, a hypothetical successor of GPT-4 trained in 2030. GPT~2030~ had a number of advanced capabi... | https://www.lesswrong.com/posts/acPYHjC9euGZRzaj6/gpt-2030-and-catastrophic-drives-four-vignettes |
# Picking Mentors For Research Programmes
Several programmes right now offer people some kind of mentor or supervisor for a few months of research. I participated in SERI MATS 4.0 over the summer, and I saw just how different people's experiences were of being mentored. So this is my list of dimensions where I think m... | https://www.lesswrong.com/posts/HzojQFXNoXjnfQvGm/picking-mentors-for-research-programmes |
# Liv Boeree Ted Talk Moloch & AI
In case you don't know, Liv Boeree is a famous professional poker player and has recently been paying attention to existential risk and coordination problems. She's part of that vague celebrity circle around rationalist diaspora that includes Tim Urban or Aella.
Since you're on LW,... | https://www.lesswrong.com/posts/trALHxmcTpckf3PNw/liv-boeree-ted-talk-moloch-and-ai |
# Update on the UK AI Summit and the UK's Plans
On November 1st and 2nd, the UK held an international AI Summit. Speeches were given, institutes founded, roundtables held, and 29 countries signed the *Bletchley Declaration.* This is a brief overview of events leading up to and at the summit, following up last month’s ... | https://www.lesswrong.com/posts/gWwMzAgDsskcb2deA/update-on-the-uk-ai-summit-and-the-uk-s-plans |
# Wastewater RNA Read Lengths
*What I'm calling "read length" here should instead have been "insert length" or "post-trimming read length". When you just say "read length" that's usually understood to be the raw length output by the sequencer, which for short-read sequencing is determined only by your target cycle cou... | https://www.lesswrong.com/posts/43EWYRcW6YcubFsrs/wastewater-rna-read-lengths |
# Artefacts generated by mode collapse in GPT-4 Turbo serve as adversarial attacks.
I stumbled upon some bizarre behaviour by GPT-4 Turbo 128k, released via API access on November 6, 2023. Since the model is in its preview phase, some imperfections are expected. Nonetheless, I found the behaviour intriguing enough to... | https://www.lesswrong.com/posts/nxhXTfsAf2LTg4xvt/artefacts-generated-by-mode-collapse-in-gpt-4-turbo-serve-as |
# Joy in the Here and Real
**Summary**: What do you do for simple, quick fun? Do you mind sharing with others?
**Tags**: Repeatable
**Purpose**: “Have fun” and “be happy” are common goals people have. If you can’t turn a few bucks and a few hours into a good time, you have a glaring gap in your abilities *and I bel... | https://www.lesswrong.com/posts/y4DtaSQdCpB5Lkxxc/joy-in-the-here-and-real |
# We have promising alignment plans with low taxes
*Epistemic status: I’m sure these plans have advantages relative to other plans. I'm not sure they're adequate to actually work, but I think they might be.*
With good enough alignment plans, we might not need coordination to survive. If alignment taxes are low enough... | https://www.lesswrong.com/posts/xqqhwbH2mq6i4iLmK/we-have-promising-alignment-plans-with-low-taxes |
# Follow-up survey: inositol
Two months ago I wrote about [inositol](https://acesounderglass.com/2023/09/08/luck-based-medicine-inositol/) as a treatment that occasionally works for anxiety and depression, especially when the user also has weird digestive issues (not medical advice, I am not a doctor). If that inspire... | https://www.lesswrong.com/posts/zQyrHb8K3zwdvMHkw/follow-up-survey-inositol |
# Linear encoding of character-level information in GPT-J token embeddings
*This post was primarily authored by MW, reporting recent research which was largely directed by suggestions from and discussions with JB. As secondary author, JB was responsible for the general shape of the post and the* [*Future Directions*](... | https://www.lesswrong.com/posts/GyaDCzsyQgc48j8t3/linear-encoding-of-character-level-information-in-gpt-j |
# Survey on the acceleration risks of our new RFPs to study LLM capabilities
[My team](https://forum.effectivealtruism.org/posts/SQSXfiByKat2YzpWu/new-roles-on-my-team-come-build-open-phil-s-technical-ai) at Open Philanthropy just launched two requests for proposals:
* Proposals to [create benchmarks](https://www.o... | https://www.lesswrong.com/posts/7qGxm2mgafEbtYHBf/survey-on-the-acceleration-risks-of-our-new-rfps-to-study |
# Who is Sam Bankman-Fried (SBF) really, and how could he have done what he did? - three theories and a lot of evidence
*For the full version of this essay, go* [*here*](https://www.spencergreenberg.com/2023/11/who-is-sam-bankman-fried-sbf-really-and-how-could-he-have-done-what-he-did-three-theories-and-a-lot-of-evide... | https://www.lesswrong.com/posts/tmsQDFerCtfBQmrDA/who-is-sam-bankman-fried-sbf-really-and-how-could-he-have |
# Memo on some neglected topics
*I originally wrote this for the *[*Meta Coordination Forum*](https://forum.effectivealtruism.org/posts/33o5jbe3WjPriGyAR/announcing-the-meta-coordination-forum-2023)*. The organizers were interested in a memo on topics *other than alignment *that might be increasingly important as AI c... | https://www.lesswrong.com/posts/3gwG5fr4FB9HC5iyK/memo-on-some-neglected-topics |
# Open Phil releases RFPs on LLM Benchmarks and Forecasting
As linked at the top of [Ajeya's "do our RFPs accelerate LLM capabilities" post](https://www.lesswrong.com/posts/7qGxm2mgafEbtYHBf/survey-on-the-acceleration-risks-of-our-new-rfps-to-study), Open Philanthropy (OP) recently released two requests for proposals ... | https://www.lesswrong.com/posts/ccNggNeBgMZFy3FRr/open-phil-releases-rfps-on-llm-benchmarks-and-forecasting |
# Palisade is hiring Research Engineers
[Palisade](https://palisaderesearch.org/) is looking to hire Research Engineers. We are a small team consisting of Jeffrey Ladish (Executive Director), Charlie Rogers-Smith (Chief of Staff), and Kyle Scott (part-time Treasurer & Operations). In joining Palisade, you would be a f... | https://www.lesswrong.com/posts/e5oXFxofxricwmH8j/palisade-is-hiring-research-engineers |
# It's OK to be biased towards humans
Let's talk about art.
In the wake of AI art generators being released, it's become pretty clear this will have a seismic effect on the art industry all across - from illustrators, to comic artists, to [animators](https://twitter.com/pika_labs/status/1722817227664306636), many cat... | https://www.lesswrong.com/posts/vBGw2raqtDHmsXQwR/it-s-ok-to-be-biased-towards-humans |
# Game Theory without Argmax [Part 1]
*Written during the SERI MATS program under the joint mentorship of* [*John Wentworth*](https://www.lesswrong.com/users/johnswentworth)*,* [*Nicholas Kees*](https://www.lesswrong.com/users/nicholaskees)*, and* [*Janus*](https://www.lesswrong.com/users/janus-1)*.*
![Create a surre... | https://www.lesswrong.com/posts/3ahqzpKvtqkom63cx/game-theory-without-argmax-part-1 |
# Game Theory without Argmax [Part 2]
*Written during the SERI MATS program under the joint mentorship of* [*John Wentworth*](https://www.lesswrong.com/users/johnswentworth)*,* [*Nicholas Kees*](https://www.lesswrong.com/users/nicholaskees)*, and* [*Janus*](https://www.lesswrong.com/users/janus-1)*.*
 (project 25: [Asymmetric control in LLMs: mo... | https://www.lesswrong.com/posts/JbmDxh5WLjX8AiTQv/control-symmetry-why-we-might-want-to-start-investigating |
# AISC project: SatisfIA – AI that satisfies without overdoing it
This project is part of the upcoming round of [AI Safety Camp](http://www.aisafety.camp/).
# SatisfIA – AI that satisfies without overdoing it
## Summary
This project will make a contribution to some fundamental design aspects of AI systems. We will... | https://www.lesswrong.com/posts/ip2Cqas89TxsWok3S/aisc-project-satisfia-ai-that-satisfies-without-overdoing-it |
# Smart Sessions - Finally a (kinda) window-centric session manager
This is a short post about some software functionality that I’ve long wanted, and a browser extension that does it well enough. I'm sharing in case anyone else has also been wanting the same thing.
### The dream
There’s a simple piece of software th... | https://www.lesswrong.com/posts/pcpNhLszE7NM4wE2v/smart-sessions-finally-a-kinda-window-centric-session |
# What ML gears do you like?
In John's [recent post](https://www.lesswrong.com/posts/nt8PmADqKMaZLZGTC/inside-views-impostor-syndrome-and-the-great-larp) he mentions many people in ML not having good gears level models of what's going on.
To wit; what gears-level models do you know for ML? How much support is there f... | https://www.lesswrong.com/posts/QoDzFtJRsRo7pqfty/what-ml-gears-do-you-like |
# The Fundamental Theorem for measurable factor spaces
I present the fundamental theorem for all finitely factored measurable spaces.
The fundamental theorem is that
two events are orthogonal if and only if they are independent in all product probability distributions.
It tells us that the definition of orthogonality ... | https://www.lesswrong.com/posts/DrchMHizfPtW3nJh2/the-fundamental-theorem-for-measurable-factor-spaces |
# Two children's stories
*(I wanted to write some rationalist-flavored bedtime stories for my friends' four-year-old. I don't really have the knack, but ChatGPT* [*did an imperfect-but-respectable job*](https://chat.openai.com/share/41ec394c-7a32-47b1-8353-70184659d597)*, given a fair bit of thematic guidance. Here ar... | https://www.lesswrong.com/posts/9CL3peBGANmA8Rjcf/two-children-s-stories |
# Zvi's Manifold Markets House Rules
This document was updated August 11, 2024. Broad intent has not changed.
All markets created by Zvi Mowshowitz shall be graded according to the rules described herein, including the zeroth rule.
The version of this on LessWrong shall be the canonical version, even if other versio... | https://www.lesswrong.com/posts/ge3Jf5Hnon8wq4xqT/zvi-s-manifold-markets-house-rules |
# You can just spontaneously call people you haven't met in years
Here's a recent conversation I had with a friend:
> Me: "I wish I had more friends. You guys are great, but I only get to hang out with you like once or twice a week. It's painful being holed up in my house the entire rest of the time."
>
> Friend: "Y... | https://www.lesswrong.com/posts/2HawAteFsnyhfYpuD/you-can-just-spontaneously-call-people-you-haven-t-met-in |
# The Science Algorithm AISC Project
I put up a project for [AISC](https://aisafety.camp/). The easiest way to evaluate your fit is to look at the [applicant questions](https://docs.google.com/document/d/1M-DMOAujWwWETkxjjAiWHRQeXXCtJm-DrAZKD6sQ5-Q/edit) I created. If the kind of reasoning required to answer these que... | https://www.lesswrong.com/posts/KHjQzxRnDCjM7xsFk/the-science-algorithm-aisc-project |
# Bostrom Goes Unheard
\[Editor’s Note: This post is split off from AI #38 and only on LessWrong because I want to avoid overloading my general readers with this sort of thing at this time, and also I think it is potentially important we have a link available. I plan to link to it from there with a short summary.\]
[... | https://www.lesswrong.com/posts/PyNqASANiAuG7GrYW/bostrom-goes-unheard |
# Research Adenda: Modelling Trajectories of Language Models
[*Apply*](https://aisafety.camp/#Projects) *to work on this project with me at AI Safety Camp 2024 before 1st December 2023.*
 *to work on this project with me at AI Safety Camp 2024 before 1st December 2023.*
Summary
=======
Future p... | https://www.lesswrong.com/posts/RHojGPWLgdFLk3PAt/aisc-project-benchmarks-for-stable-reflectivity |
# Suggestions for chess puzzles
(You can sign up [here](https://docs.google.com/forms/d/e/1FAIpQLScPKrSB6ytJcXlLhnxgvRv1V4vMx8DXWg1j9KYVfVT1ofdD-A/viewform?vc=0&c=0&w=1&flr=0) if you haven't already.)
As [suggested](https://www.lesswrong.com/posts/ddsjqwbJhD9dtQqDH/lying-to-chess-players-for-alignment?commentId=dfojS... | https://www.lesswrong.com/posts/SuGw2E2z68SfAMkAt/suggestions-for-chess-puzzles |
# Non-myopia stories
*Written under the supervision of Lionel Levine. Thanks to Owain Evans, Aidan O’Gara, Max Kaufmann, and Johannes Treutlein for comments.*
This post is a synthesis of arguments made by other people. It provides a collection of answers to the question, "Why would an AI become non-myopic?" In this p... | https://www.lesswrong.com/posts/2ZnvFDtSWc3Hteah6/non-myopia-stories |
# How to Upload a Mind (In Three Not-So-Easy Steps)
[Cross-posted to the EA forum](https://forum.effectivealtruism.org/posts/BLPaNx6LhPBZxDsSM/how-to-upload-a-mind-in-three-not-so-easy-steps)
*This Rational Animations video is about the research and practical challenges of "whole brain emulation" or "mind uploading",... | https://www.lesswrong.com/posts/PnBFLWiX5p36CJyTH/how-to-upload-a-mind-in-three-not-so-easy-steps |
# They are made of repeating patterns
*Epistemis status: an obvious parody. *
— You won't believe me. I've found them.
— Whom?
— Remember that famous discovery by Professor Prgh'zhyne about pockets of baryonic matter in open systems that minimize the production of entropy within them? They went further and claimed ... | https://www.lesswrong.com/posts/xCPcn8cjjeC6PnB5z/they-are-made-of-repeating-patterns |
# Theories of Change for AI Auditing
Executive summary
=================
Our mission at Apollo Research is to reduce catastrophic risks from AI by auditing advanced AI systems for misalignment and dangerous capabilities, with an initial focus on deceptive alignment.
In our [announcement post](https://www.alignmentfo... | https://www.lesswrong.com/posts/LwJwDNFhjurAKFiJm/theories-of-change-for-ai-auditing |
# Great Empathy and Great Response Ability
As kids, long before our frontal lobes have finished developing, we’ve all had the instinct to tease. Just watch two toddlers interact, you can see the malicious glint in their eye as they discover that calling one of their peers a “poopy head” triggers a devastating emotiona... | https://www.lesswrong.com/posts/eKNBzq78i7dToHanu/great-empathy-and-great-response-ability |
# Loudly Give Up, Don't Quietly Fade
### I.
There’s a supercharged, dire wolf form of the bystander effect that I’d like to shine a spotlight on.
First, a quick recap. The Bystander Effect is a phenomenon where people are less likely to help when there’s a group around. When I took basic medical training, I was told... | https://www.lesswrong.com/posts/bkfgTSHhm3mqxgTmw/loudly-give-up-don-t-quietly-fade |
# What is wisdom?
# Laterally through the chronophone
In "[Archimedes's Chronophone](https://www.lesswrong.com/posts/cKrgy7hLdszkse2pq/archimedes-s-chronophone)", Yudkowsky asks: What would you say to Archimedes——what important message would you want to send back in time, to set the world on a hopeworthy course ... | https://www.lesswrong.com/posts/fzKfzXWEBaENJXDGP/what-is-wisdom-1 |
# Is there Work on Embedded Agency in Cellular Automata Toy Models?
Is there any work that is about modeling embedded agency in a cellular automata? That seems like an obvious thing to do. I am asking in general and specifically in the context of solving embedded agency using [these assumptions](https://www.lesswrong.... | https://www.lesswrong.com/posts/ryhzvdgHEH77QTzcb/is-there-work-on-embedded-agency-in-cellular-automata-toy |
# Do you want a first-principled preparedness guide to prepare yourself and loved ones for potential catastrophes?
Many of us have heard how we should prepare for potential disasters. But is such advice outdated given increasing threats from both AI and biotechnology? This post is both a test for people’s appetite for... | https://www.lesswrong.com/posts/LtshicMv63iybgTdi/do-you-want-a-first-principled-preparedness-guide-to-prepare |
# Monthly Roundup #12: November 2023
Things on the AI front have been rather hectic. That does not mean other things stopped happening. Quite the opposite. So here we are again.
#### Bad News
[PSA: Crumbl Cookies, while delicious, have rather a lot of calories](https://twitter.com/tolstoybb/status/171213455266641133... | https://www.lesswrong.com/posts/TfABomJ7s6xLkxTFz/monthly-roundup-12-november-2023 |
# More metal less ore
Epistemic status: musings on advice from three different spaces, all with a similar tactic to improve your life. This discovery is told through a semi-fictional story.
I am not the happiest person.
Rather than accepting this, I research it, I make it my job to change that fact.
My research br... | https://www.lesswrong.com/posts/wxsLeekp6zwWSuAPL/more-metal-less-ore |
# Raemon's Deliberate (“Purposeful?”) Practice Club
Introduction
============
So. I have a theory of [feedbackloop-first rationality.](https://www.lesswrong.com/posts/pZrvkZzL2JnbRgEBC/feedbackloop-first-rationality)
It has a lot of parts. I think each part is promising on it's own, and I have a dream that they int... | https://www.lesswrong.com/posts/9tx4jRAuEddap7Tzp/raemon-s-deliberate-purposeful-practice-club |
# Kids or No kids
*This post summarizes how my partner and I decided whether to have children or not. We spent hundreds of hours on this decision and hope to save others part of that time. We found it very useful to read the thoughts of people who share significant parts of our values on the topic and thus want to "pa... | https://www.lesswrong.com/posts/3MzDMBk4DZrbYePJS/kids-or-no-kids |
# Facebook is Paying Me to Post
In mid-September Facebook asked me if I wanted to start getting paid for posting things:
[](https://www.jefftk.com/performance-bonus-program-offer-big.png)
This seems to be the [performance bonus program](https://www.faceb... | https://www.lesswrong.com/posts/hbyEjaz2syZjDB4us/facebook-is-paying-me-to-post |
# Betting on what is un-falsifiable and un-verifiable
This post is part of my broader aim to try and formulate everything AI in terms of markets so that alignment becomes an antitrust problem. Comments appreciated.
Prerequisites[^6yhfqi0t6ie].
This post is also available as a (more obtusely-written) [arXiv paper](h... | https://www.lesswrong.com/posts/id84oe3LxdzoqinKA/betting-on-what-is-un-falsifiable-and-un-verifiable |
# When did Eliezer Yudkowsky change his mind about neural networks?
In 2008, Eliezer Yudkowsky was strongly critical of neural networks. From his post "[Logical or Connectionist AI?](https://www.lesswrong.com/posts/juomoqiNzeAuq4JMm/logical-or-connectionist-ai)":
> Not to mention that neural networks have also been "... | https://www.lesswrong.com/posts/WyJKqCNiT7HJ6cHRB/when-did-eliezer-yudkowsky-change-his-mind-about-neural |
# Linearity Fallacies
A linearity fallacy is committed when you incorrectly assume two variables are related by a linear relationship. For example, assuming that adding people to a project decreases the time to finish proportionally[^d1ra3dfyry] or that happiness scales linearly with income. This kind of fallacy, wher... | https://www.lesswrong.com/posts/dgvTCGS5egCdZ6uJF/linearity-fallacies-1 |
# LLMs May Find It Hard to FOOM
*Epistemic status: some of the technological progress parts of this I've been thinking about for many years, other more LLM-specific parts I have been thinking about for months or days.*
**TL;DR** An extremely high capacity LLM trained on extremely large amounts content from humans wil... | https://www.lesswrong.com/posts/qGTxGGNxcciY2nrHv/llms-may-find-it-hard-to-foom |
# Incidental polysemanticity
$%%% GENERIC MATH %%%
%
% Delimiters
\newcommand{\p}[1]{(#1)}
\newcommand{\ps}[1]{\left(#1\right)}
\newcommand{\P}[1]{\!\ps{#1}}
\newcommand{\norm}[1]{\|#1\|}
\newcommand{\abs}[1]{|#1|}
\newcommand{\coloneqq}{:=}
\newcommand{\eqqcolon}{=:}
%
% Shortcuts
\newcommand{\al}[1]{\begin{align}#1\... | https://www.lesswrong.com/posts/sEyWufriufTnBKnTG/incidental-polysemanticity |
# Reinforcement Via Giving People Cookies
I.
--
[Thinking By The Clock](https://www.lesswrong.com/posts/WJtq4DoyT9ovPyHjH/thinking-by-the-clock) is now the most popular thing I've written on LessWrong, so here's another entry in the list of things which had a significant change in how I think and operate that I learn... | https://www.lesswrong.com/posts/wFGEtyHvn699dNshg/reinforcement-via-giving-people-cookies |
# A conceptual precursor to today's language machines [Shannon]
*Cross-posted* [*from New Savanna*](A conceptual precursor to today's language machines [Shannon]).
I'm in the process of reading a fascinating article by Richard Hughes Gibson, [Language Machinery: Who will attend to the machines’ writing?](https://hedg... | https://www.lesswrong.com/posts/Y7WP47tL9zQwkLTqZ/a-conceptual-precursor-to-today-s-language-machines-shannon |
# Experiences and learnings from both sides of the AI safety job market
*I’m writing this in my own capacity. The views expressed are my own, and should not be taken to represent the views of Apollo Research or any other program I’m involved with. *
In 2022, I applied to multiple full-time AI safety positions. Now, I... | https://www.lesswrong.com/posts/WqYSmjSsE3hi8Lgot/experiences-and-learnings-from-both-sides-of-the-ai-safety |
# 'Theories of Values' and 'Theories of Agents': confusions, musings and desiderata
Meta:
* *Content signposts:* we talk about limits to expected utility theory; what values are (and ways in which we're confused about what values are); the need for a "generative"/developmental logic of agents (and their values); ty... | https://www.lesswrong.com/posts/mbpMvuaLv4qNEWyG6/theories-of-values-and-theories-of-agents-confusions-musings |
# AISN #26: National Institutions for AI Safety, Results From the UK Summit, and New Releases From OpenAI and xAI
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.... | https://www.lesswrong.com/posts/HWudwSfKLeAfg8Co6/aisn-26-national-institutions-for-ai-safety-results-from-the |
# Large Language Models can Strategically Deceive their Users when Put Under Pressure.
Results of an autonomous stock trading agent in a realistic, simulated environment.
> We demonstrate a situation in which Large Language Models, trained to be helpful, harmless, and honest, can display misaligned behavior and strat... | https://www.lesswrong.com/posts/4mM8RYsm4okrqGSqx/large-language-models-can-strategically-deceive-their-users |
# New report: "Scheming AIs: Will AIs fake alignment during training in order to get power?"
(Cross-posted from [my website](https://joecarlsmith.com/2023/11/15/new-report-scheming-ais-will-ais-fake-alignment-during-training-in-order-to-get-power))
I’ve written a report about whether advanced AIs will fake alignment ... | https://www.lesswrong.com/posts/yFofRxg7RRQYCcwFA/new-report-scheming-ais-will-ais-fake-alignment-during |
# Testbed evals: evaluating AI safety even when it can’t be directly measured
A few collaborators and I recently released the paper, “Generalization Analogies (GENIES): A Testbed for Generalizing AI Oversight to Hard-To-Measure Domains. ([tweet thread](https://twitter.com/joshua_clymer/status/1724851456967417872?s=2... | https://www.lesswrong.com/posts/okmB8ymyhgc65WckN/testbed-evals-evaluating-ai-safety-even-when-it-can-t-be |
# Glomarization FAQ
The main reason I'm posting this here is because I want it to be publicly recorded that I wrote this post in 2023. That way, I can show people in the future that I didn't just make this up on the spot; it's been around since long before whatever it is they're asking me about this time. I don't have... | https://www.lesswrong.com/posts/N5txrJDimv8nz4n24/glomarization-faq |
# Life on the Grid (Part 1)
The physical layout of the environment in which someone is raised impacts their cognitive abilities as an adult. [A recent study](https://www.nature.com/articles/s41586-022-04486-7) based on video game data from 397,162 people across 38 countries found that “people who grew up outside citie... | https://www.lesswrong.com/posts/j4cKyhDEBpGPLNpig/life-on-the-grid-part-1 |
# In Defense of Parselmouths
Prerequisites: [The Quaker and the Parselmouth](http://benjaminrosshoffman.com/the-quaker-and-the-parselmouth/).
I.
--
First, a quick summary.
In the prerequisite post, Benjamin Hoffman describes three kinds of people. These people are hypothetical extremes: they're the social and epist... | https://www.lesswrong.com/posts/R28YGeAzDHehrnc7f/in-defense-of-parselmouths |
# Extrapolating from Five Words
If you only get [about five words](https://www.lesswrong.com/posts/4ZvJab25tDebB8FGE/you-get-about-five-words) to convey an idea, what will someone extrapolate from those five words? Rather than guess, you can use LLMs to experimentally discover what people are likely think those five w... | https://www.lesswrong.com/posts/HD8kLyYcSuYRi4vzP/extrapolating-from-five-words |
# Learning coefficient estimation: the details
What this is for
----------------
The learning coefficient (LC), or RLCT, is a quantity from singular learning theory that can help to quantify the "complexity" of deep learning models, among other things.
This guide is primarily intended to help people interested in im... | https://www.lesswrong.com/posts/9ecpBaAiGQnkmX9Ex/learning-coefficient-estimation-the-details |
# How much fraud is there in academia?
Quinn [wrote](https://www.lesswrong.com/posts/LizpBcsF9XEAhTsvn/linkpost-they-studied-dishonesty-was-their-work-a-lie?commentId=XoKFKW55yyg3gsEcc) a while ago "I heard a pretty haunting take about how long it took to discover steroids in bike races. Apparently, there was a while ... | https://www.lesswrong.com/posts/PyshEEKcxTLdeMgfm/how-much-fraud-is-there-in-academia |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.