text stringlengths 300 320k | source stringlengths 52 154 |
|---|---|
# EA & LW Forums Weekly Summary (24 - 30th Oct 22')
*Supported by Rethink Priorities*
This is part of a weekly series - you can see the full collection [here.](https://forum.effectivealtruism.org/s/W4fhpuN26naxGCBbN) The first post includes some details on purpose and methodology.
If you'd like to receive these summ... | https://www.lesswrong.com/posts/soCpjCJYhAcDaoZWu/ea-and-lw-forums-weekly-summary-24-30th-oct-22 |
# ML Safety Scholars Summer 2022 Retrospective
**TLDR**
========
This is a report on the [Machine Learning Safety Scholars program](https://forum.effectivealtruism.org/posts/9RYvJu2iNJMXgWCBn/introducing-the-ml-safety-scholars-program), organized over the summer by the [Center for AI Safety](https://safe.ai). 63 stud... | https://www.lesswrong.com/posts/LykaymHy5KLhr6JXA/ml-safety-scholars-summer-2022-retrospective |
# Conversations on Alcohol Consumption
Recently I have decided to stop drinking alcohol. Below are some conversations with two friends who are on similar paths. I decided to post these conversations because I am looking for as many perspectives as possible and the LessWrong community might have some insights. I will k... | https://www.lesswrong.com/posts/4LBqvkkSPH27Qr2dY/conversations-on-alcohol-consumption |
# Remember to translate your thoughts back again
***Epistemic status:** casual observation written as first draft but published as-is because it was unlikely to ever actually get edited. *
There's a common failure mode I see among EAs and particularly rationalists where they fail to translate their thoughts back when... | https://www.lesswrong.com/posts/ztdSy3FRcJ6zwqCeu/remember-to-translate-your-thoughts-back-again |
# Auditing games for high-level interpretability
*This post was written under Evan Hubinger’s direct guidance and mentorship, as a part of the *[*Stanford Existential Risks Institute ML Alignment Theory Scholars (MATS) program*](https://www.lesswrong.com/posts/8vLvpxzpc6ntfBWNo/seri-ml-alignment-theory-scholars-progra... | https://www.lesswrong.com/posts/EbL5W5ccwfbqFiYBJ/auditing-games-for-high-level-interpretability-1 |
# Clarifying AI X-risk
*TL;DR: We give a threat model literature review, propose a categorization and describe a consensus threat model from some of DeepMind's AGI safety team. See our* [*post*](https://www.alignmentforum.org/posts/wnnkD6P2k2TfHnNmt/threat-model-literature-review) *for the detailed literature review.*... | https://www.lesswrong.com/posts/GctJD5oCDRxCspEaZ/clarifying-ai-x-risk |
# Threat Model Literature Review
*TL;DR: This post provides a literature review of some threat models of how misaligned AI can lead to existential catastrophe. See our accompanying* [*post*](https://www.alignmentforum.org/posts/GctJD5oCDRxCspEaZ/clarifying-ai-x-risk) *for high-level discussion, a categorization and ou... | https://www.lesswrong.com/posts/wnnkD6P2k2TfHnNmt/threat-model-literature-review |
# Progress links and tweets, 2022-11-01
**Announcements**
-----------------
* [The Reef Starter Innovation Challenge, to identify startups that could benefit from operations in Low Earth Orbit](https://reefstarterchallenge.techconnectventures.com/)
* [AI startup Generally Intelligent has launched and open-sourced... | https://www.lesswrong.com/posts/JAgPXWoCMJvxEpWWW/progress-links-and-tweets-2022-11-01 |
# Mildly Against Donor Lotteries
Let's say you're going to donate some money, and plan to put some time into figuring out the best place to donate. The more time you put into your decision the better it's likely to be, but at some point you need to stop looking and actually make the donation. The more you're donating ... | https://www.lesswrong.com/posts/R2n8gw7k2maZtk7Bk/mildly-against-donor-lotteries |
# Open & Welcome Thread - November 2022
If it’s worth saying, but not worth its own post, here's a place to put it.
If you are new to LessWrong, here's the place to introduce yourself. Personal stories, anecdotes, or just general comments on how you found us and what you hope to get from the site and community are in... | https://www.lesswrong.com/posts/WCHKpqvdpGhiFtLTc/open-and-welcome-thread-november-2022 |
# All AGI Safety questions welcome (especially basic ones) [~monthly thread]
**tl;dr: Ask questions about AGI Safety as comments on this post, including ones you might otherwise worry seem dumb!**
Asking beginner-level questions can be intimidating, but everyone starts out not knowing anything. If we want more people... | https://www.lesswrong.com/posts/fSMrwJnqRb5NrMYFx/all-agi-safety-questions-welcome-especially-basic-ones-2 |
# Real-Time Research Recording: Can a Transformer Re-Derive Positional Info?
**New experiment:** Recording myself real-time as I do mechanistic interpretability research! I try to answer the question of what happens if you train a toy transformer without positional embeddings on the task of "predict the previous token... | https://www.lesswrong.com/posts/sYHrW4wwfoMBxNDcA/real-time-research-recording-can-a-transformer-re-derive |
# Sequence Reread: Fake Beliefs [plus sequence spotlight meta]
A couple months ago the LessWrong team started putting Sequence Spotlights at the top of the home page, to give people reasonably bite-sized chunks of canonical LessWrong content to read (or reread). i.e, this thing here:
 as the parts of plants that we can eat but not digest. We're always told to eat more of it. Why should we eat things we can't digest?
There are many papers to point to... | https://www.lesswrong.com/posts/PKcpb9BEonNmRNv2w/why-is-fiber-good-for-you |
# AI Safety Needs Great Product Builders
In his [AI Safety Needs Great Engineers](https://forum.effectivealtruism.org/posts/DDDyTvuZxoKStm92M/ai-safety-needs-great-engineers) post, Andy Jones explains how software engineers can reduce the risks of unfriendly artificial intelligence. Even without deep ML knowledge, the... | https://www.lesswrong.com/posts/ouuhffhvvBkTfzxTY/ai-safety-needs-great-product-builders |
# Mind is uncountable
I'd like to share a simple thought experiment that has some pretty strange implications. It seems to show that the world cannot always be sharply divided into separate experiences.
We can make the analogy, to how the world cannot always be divided into separate organisms. Usually it can, but the... | https://www.lesswrong.com/posts/MJ9GdeSXyR4p9xvHp/mind-is-uncountable |
# Housing and Transit Thoughts #1
Housing is the Balsa-style area where there has been the most progress lately. There seems to be growing consensus that it is vital we build more housing in places that people want to live. California is making serious attempts to force localities to do exactly that, here in New York ... | https://www.lesswrong.com/posts/4EDnZwLDzHhA3CFY7/housing-and-transit-thoughts-1 |
# Far-UVC Light Update: No, LEDs are not around the corner (tweetstorm)
I wrote a [tweetstorm](https://twitter.com/davidmanheim/status/1587757208120557568) on why 222nm LEDs are not around the corner, and given that there has been [some](https://www.lesswrong.com/posts/3yCcdg7DQQnfqZoax/far-ultraviolet-light-in-public... | https://www.lesswrong.com/posts/4Zjm8ycWhg6PzFwnF/far-uvc-light-update-no-leds-are-not-around-the-corner |
# Humans do acausal coordination all the time
I used to think that acausal coordination was a weird thing that AI’s might do in the future, but that they certainly wouldn’t learn from looking at human behavior. I don’t believe that anymore, and think there are lots of examples of acausal coordination in everyday life.... | https://www.lesswrong.com/posts/FCffGHJnYfdE2DgRe/humans-do-acausal-coordination-all-the-time |
# Is there a good way to award a fixed prize in a prediction contest?
Let's say we're having a prediction contest where we want people to pre-register their probability estimates for the outcomes of a set of discrete events (e.g. the candidates for various races in an upcoming election). We have a fixed sum of, say, $... | https://www.lesswrong.com/posts/EjYRGqe9QhgWyoQAh/is-there-a-good-way-to-award-a-fixed-prize-in-a-prediction |
# Lazy Python Argument Parsing
When I'm writing real Python programs I use [argparse](https://docs.python.org/3/library/argparse.html). When I'm writing quick scripts, I destructure `sys.argv[1:]`. For example:
width, depth, height = sys.argv\[1:\]
...
Or, if I need to `import` the file:
def start(width, depth, hei... | https://www.lesswrong.com/posts/GzF7JufwBMgM66qct/lazy-python-argument-parsing |
# Open Letter Against Reckless Nuclear Escalation and Use
There's been ample interest in the probability that the Ukraine conflict escalates into a nuclear way, ranging from predictions on [Metaculus](https://www.metaculus.com/questions/12591/nuclear-detonation-in-ukraine-by-2023/), [Manifold](https://manifold.markets... | https://www.lesswrong.com/posts/hRqop6A6ho2dyhpSC/open-letter-against-reckless-nuclear-escalation-and-use |
# The Mirror Chamber: A short story exploring the anthropic measure function and why it can matter
*This story warrants and motivates the study of the hard problem of consciousness, presenting a plausible future scenario in which a notion of quantifiable anthropic measure as a property of an arrangement of matter **de... | https://www.lesswrong.com/posts/9RdhJKPrYvsttsko9/the-mirror-chamber-a-short-story-exploring-the-anthropic |
# K-types vs T-types — what priors do you have?
**Summary:** There is a spectrum between two types of people, K-types and T-types. K-types want theories with low kolmogorov-complexity and T-types want theories with low time-complexity. This classification correlates with other classifications and with certain personal... | https://www.lesswrong.com/posts/xxrkDsBHJ8MuCNWjf/k-types-vs-t-types-what-priors-do-you-have |
# Covid 11/3/22: Asking Forgiveness
It was a quiet week on the Covid front. That is a very good thing.
If anything, the theme of this week was _looking backwards_ to how things were handled. The CDC is (at least half-heartedly) looking to fix things. There was an offer of ‘pandemic amnesty’ by someone who thinks they... | https://www.lesswrong.com/posts/d7PMSkruxK3xpj52x/covid-11-3-22-asking-forgiveness |
# Multiple Deploy-Key Repos
Let's say you want to give a machine access to one of your repos but not all of them. This is a reasonable thing to want, and so GitHub offers [deploy keys](https://docs.github.com/en/developers/overview/managing-deploy-keys). But what if you want to give it access to two repos?
The natura... | https://www.lesswrong.com/posts/ufaa6F72BikAoAmsd/multiple-deploy-key-repos |
# A Mystery About High Dimensional Concept Encoding
*Epistemic status: I ran some experiments with surprising results, and I don’t have a clear intuition of what it means for the way concepts are encoded in model activations. Help is welcome!*
***\[EDIT\] I made a mistake while analyzing the data, and it changes some... | https://www.lesswrong.com/posts/QHKfYy9LLAsjC5rTK/a-mystery-about-high-dimensional-concept-encoding |
# [Video] How having Fast Fourier Transforms sooner could have helped with Nuclear Disarmament - Veritasium
Veritasium presents an argument that if we'd had Fast Fourier Transforms sooner, this would have made it possible to detect underground nuclear tests without on-site audits, somewhat better nuclear non-prolifera... | https://www.lesswrong.com/posts/YbzguicgvSEkYe3zs/video-how-having-fast-fourier-transforms-sooner-could-have |
# Could a Supreme Court suit work to solve NEPA problems?
According to [Eli Dourado](https://www.thecgo.org/benchmark/much-more-than-you-ever-wanted-to-know-about-nepa/), the original interpretation of the NEPA law was reasonable. The earliest Environmental Impact Statements (EIS) were often less than ten pages. After... | https://www.lesswrong.com/posts/6Hz3DKt6mqq5gHDD3/could-a-supreme-court-suit-work-to-solve-nepa-problems |
# Mechanistic Interpretability as Reverse Engineering (follow-up to "cars and elephants")
I think (perhaps) the distinction that I was trying to make in my previous post ["Cars and Elephants": a handwavy argument/analogy against mechanistic interpretability](https://www.lesswrong.com/posts/YEkzeJTrp69DTn8KD/cars-and-e... | https://www.lesswrong.com/posts/kjRGMdRxXb9c5bWq5/mechanistic-interpretability-as-reverse-engineering-follow |
# Are alignment researchers devoting enough time to improving their research capacity?
(Epistemic Status: Anecdotal)
If we want to reduce AGI x-risk, it seems pretty intuitive to me that alignment researchers should be regularly dedicating time to improving their research capacity. But I'm suspicious that many of... | https://www.lesswrong.com/posts/9cKf2BBR4X2JTSeiz/are-alignment-researchers-devoting-enough-time-to-improving |
# A newcomer’s guide to the technical AI safety field
*This post was written during *[*Refine*](https://www.lesswrong.com/posts/5uiQkyKdejX3aEHLM/how-to-diversify-conceptual-alignment-the-model-behind)*. Thanks to Jonathan Low, Linda Linsefors, Koen Holtman, Aaron Scher, and Nicholas Kees Dupuis for helpful discussion... | https://www.lesswrong.com/posts/5rsa37pBjo4Cf9fkE/a-newcomer-s-guide-to-the-technical-ai-safety-field |
# Weekly Roundup #4
Relatively light roundup this week, likely due to Elon Musk buying Twitter plus the incoming midterms sucking all the oxygen from everything else. I do have a Twitter/Musk post in progress tentatively scheduled for Monday.
### Prediction Markets Would Probably Be Great
Journalist Ben Collins says... | https://www.lesswrong.com/posts/yguz9d8BzvWeLXpm8/weekly-roundup-4 |
# Monthly Shorts 10/22
### India?
Map of Indian poverty, raw data on page [136](https://web.archive.org/web/20220131074736/https://www.niti.gov.in/sites/default/files/2021-11/National_MPI_India-11242021.pdf).
[
 has been really gratifying. Before I published I was quite afraid to talk about my emotional response to the problem, and worried that people would strong arm in the comments. The former didn’t happen and the latter ... | https://www.lesswrong.com/posts/fubHaLebjJRnbEkHc/follow-up-to-medical-miracle |
# Toy Models and Tegum Products
(*Thanks to Adam Scherlis, Kshitij Sachan, Buck Shlegeris, Chris Olah, and Nicholas Schiefer for conversations that informed this post. Thanks to Aryan Bhatt for catching an error in the loss minimization.)*
Anthropic recently published a paper on toy models of superposition \[[Elhage+... | https://www.lesswrong.com/posts/3cR2YH9dpr7SmKvCb/toy-models-and-tegum-products |
# For ELK truth is mostly a distraction
*Epistemic Status: Pretty confident in the central conclusions, and very confident in the supporting claims from meta-logic. Any low confidence conclusions are presented as such. NB: I give an intentionally revisionary reading of what ELK is (or should be) about. Accordingly, I ... | https://www.lesswrong.com/posts/NxApPkbjt9hXraSts/for-elk-truth-is-mostly-a-distraction |
# How much should we care about non-human animals?
This is inspired by a [long and passionate post](https://www.lesswrong.com/posts/hRqop6A6ho2dyhpSC/open-letter-against-reckless-nuclear-escalation-and-use?commentId=AEJwJfvWF2MSpFiYP) from [Bernd Clemens Huber](https://www.lesswrong.com/users/bernd-clemens-huber). It... | https://www.lesswrong.com/posts/j3gPzDMLjnxzgJHQR/how-much-should-we-care-about-non-human-animals |
# Should we “go against nature”?
Should we “go against nature”? Or live in “harmony” with it? There are two senses of “nature.” Teasing them apart clarifies this issue.
“Nature” can mean immutable natural law. We defy this at our peril. If we dump raw sewage where we get our drinking water, we will suffer epidemics. ... | https://www.lesswrong.com/posts/TrGhJ89ihj5hBDyiz/should-we-go-against-nature |
# Metaculus is seeking Software Engineers
(See Metaculus's previous career-tagged posts [here](https://www.lesswrong.com/posts/bGueGuSsDpRPJF8aD/metaculus-is-seeking-analytical-storytellers-to-write-essays), [here,](https://www.lesswrong.com/posts/WaEd8oHHKD5AcYzBu/metaculus-is-seeking-experienced-leaders-researchers-... | https://www.lesswrong.com/posts/nu3xb9mHWceJwwSHA/metaculus-is-seeking-software-engineers |
# Spectrum of Independence
When I compare our parenting decisions to peers and neighbors' we tend to be near the far end of thinking our kids are ready to do things [on their own](https://www.jefftk.com/p/growing-independence). I don't see other first graders (6y) unaccompanied at the park or [crossing the street](htt... | https://www.lesswrong.com/posts/rdidWGxzvZDF5zYCS/spectrum-of-independence |
# The Slippery Slope from DALLE-2 to Deepfake Anarchy
OpenAI developed DALLE-2 to be safe. Then StabilityAI made an open-source copycat.
===================================================================================
Stephen Casper ([scasper@mit.edu](mailto:scasper@mit.edu))
Phillip Christoffersen ([philljkc@m... | https://www.lesswrong.com/posts/BeGCCmDtkdJD7j5Y5/the-slippery-slope-from-dalle-2-to-deepfake-anarchy |
# Provably Honest - A First Step
Introduction
============
I am new to AI Safety and this is my first lesswrong post (crossposted to EA Forum) so please feel free to correct my mistakes. This post is going to be a bit about my interpretation of existing alignment research and how I intend to tackle the problem, and t... | https://www.lesswrong.com/posts/d3fgJb3RED258TciH/provably-honest-a-first-step |
# Instead of technical research, more people should focus on buying time
*This post is the first in a sequence of posts about AI strategy. In the next post, we’ll provide more examples of “buying time” interventions that we’re excited about. *
*We’re grateful to Ajeya Cotra, Daniel Kokotajlo, Ashwin Acharya, and Andr... | https://www.lesswrong.com/posts/BbM47qBPzdSRruY4z/instead-of-technical-research-more-people-should-focus-on |
# Recommend HAIST resources for assessing the value of RLHF-related alignment research
**Overview:** we're trying to form inside views about the value of alignment research related to RLHF. Recommend us reading material expressing relevant arguments.
The [Harvard AI Safety Team](https://haist.ai/) (HAIST) is the AI s... | https://www.lesswrong.com/posts/ehK7WtBsDfiCzXTw8/recommend-haist-resources-for-assessing-the-value-of-rlhf |
# Unpricable Information and Certificate Hell
**Certificate**: Expensive Shareable evidence of some desirable quality, typically of a marketable good. Increases the percieved value of that good to the market and so benefits the owner, but has a negative externality of lowering the percieved value of all similar uncert... | https://www.lesswrong.com/posts/bb5ePG4i4m3sPw2GT/unpricable-information-and-certificate-hell |
# Takeaways from a survey on AI alignment resources
What am I talking about?
------------------------
In June and July of this year, I ran a survey to ask a lot of people how useful they found a variety of resources on AI alignment. I was particularly interested in “secondary resources”: that is, not primary resource... | https://www.lesswrong.com/posts/rXSBvSKvKdaNkhLeJ/takeaways-from-a-survey-on-ai-alignment-resources |
# Has anyone increased their AGI timelines?
The question: Within the last 5-10 years, is there is any person or group that has openly increased their AGI timelines?
Ideally, they would have at least two different estimates (years apart?), with the most recent estimate showing that they think AGI is further into the... | https://www.lesswrong.com/posts/hr48gem2keDQvEAbg/has-anyone-increased-their-agi-timelines |
# Response
*Apotheosis, Home, Works*
* * *
**Part V**
Everyone has their own name for it.
"[The Multi-Dimensional Crisis Revolution"](http://metamoderna.org/the-boom-equation/?lang=en), "[The Precipice"](https://theprecipice.com/), "[Collapse](https://mobile.twitter.com/anderssandberg/status/1495388640238780418)",... | https://www.lesswrong.com/posts/2LprYDWQuqSmYhoQ5/response |
# What is epigenetics?
Among all areas of biology related to my research, epigenetics is the one that is most commonly misunderstood, not only by the general public but even by other scientists. After being irritated one too many times,[^odwminrhfna] I’ve decided to make a series of posts to explain what epigenetics r... | https://www.lesswrong.com/posts/NaEAuehWLvuRSTzHm/what-is-epigenetics |
# Apple Cider Syrup
We bought some apple cider when we went apple picking a few weeks ago, and while it's tasty stuff it seems it's not what people wanted to be [spending their sweet budget on](https://www.jefftk.com/p/how-much-is-a-sweet): we still had 2/3 of it left three weeks later. I decided to reduce it and make... | https://www.lesswrong.com/posts/nYtW9wJkRCBSaBt8u/apple-cider-syrup |
# You won’t solve alignment without agent foundations
The following is a part of a response to a Vivek Hebbar's problem (the questions in *cursive*). I looked at why an alignment proposal doesn't work at an AGI level, in connection to only two of [Eliezer's doom arguments](https://www.lesswrong.com/posts/uMQ3cqWDPHhjt... | https://www.lesswrong.com/posts/3dFogxGK8uNv5xCSv/you-won-t-solve-alignment-without-agent-foundations |
# Playground Game: Monster
Here's a game I play with the kids at the park a lot:
* You play on a play structure.
* One person is the "monster"
* Anyone they tag becomes a monster as well.
* Monsters are not allowed on the horizontal surfaces of the play structure.
* Non-monsters can't go... | https://www.lesswrong.com/posts/iyDbA7jvabyLBi3pD/playground-game-monster |
# Exams-Only Universities
**Quality**: fast write-up (~45 minutes); little researched
I want a university that only does exams, which would include a description of what you need to know for the exams. Bonus would be suggestions for textbooks and online classes to learn the material, but that's optional.
# Cost
The ... | https://www.lesswrong.com/posts/vH7n8tuPxpPyKJpuX/exams-only-universities |
# Is there any discussion on avoiding being Dutch-booked or otherwise taken advantage of one's bounded rationality by refusing to engage?
From [https://www.gwern.net/mugging:](https://www.gwern.net/mugging:)
> One way to try to escape a mugging is to unilaterally declare that all probabilities below a certain small p... | https://www.lesswrong.com/posts/6XsZi9aFWdds8bWsy/is-there-any-discussion-on-avoiding-being-dutch-booked-or |
# A philosopher's critique of RLHF
In the spring, I went to a talk with Brian Christian at Yale. He talked about his book, *The Alignment Problem*, and then there was an audience Q&A. There was a really remarkable question in that Q&A, which I have transcribed here. It came from the Yale philosophy professor [L.A. Pau... | https://www.lesswrong.com/posts/scnkAbvLMDjJR9WE2/a-philosopher-s-critique-of-rlhf |
# 2022 LessWrong Census?
From 2011-2017, there was an annual LessWrong census/survey. Much like a national census, this provided a valuable lens into the demographics and beliefs of LessWrongers. Unfortunately, this tradition appears to have stopped in recent years, with the exception of a mini-revival in [2020](https... | https://www.lesswrong.com/posts/jokDcnRP5cpFWjp8y/2022-lesswrong-census |
# Counterfactability
This post will assume an understanding of the [finite factored set](https://www.lesswrong.com/s/kxs3eeEti9ouwWFzr) ontology. It will be more speculative that the main FFS sequence, and I will leave out some proofs. It seems likely that I will later regret some of the definitions laid out here. I w... | https://www.lesswrong.com/posts/b2YBddoCKSixivSAJ/counterfactability |
# [Book Review] "Station Eleven" by Emily St. John Mandel
# TL;DR
Is it good? Yes.
Should you read it? I don't know.
# Spoilers Ahead
*Station Eleven* isn't the kind of book I'd normally read but I wanted to go to a book club and *Station Eleven* is what they were reading so I bought a copy and quickly read it ove... | https://www.lesswrong.com/posts/YdZDLxeKcB5wGXLE2/book-review-station-eleven-by-emily-st-john-mandel |
# Google Search as a Washed Up Service Dog: "I HALP!"
As is customary, I whined (on IRC) about Google Search being unhelpful or actively counterproductive with the results these days and how it is not wh... | https://www.lesswrong.com/posts/naAs59xiGfr7fPjej/google-search-as-a-washed-up-service-dog-i-halp |
# 4 Key Assumptions in AI Safety
AGI is already a very difficult topic, with a large amount of uncertainty, and not enough research being done for its core problems. However, I think even with this understanding, its vastness in scope of possibilities might still be severely under-realized. I’ve listed four major assu... | https://www.lesswrong.com/posts/5KAhnDbq9F4a2Y2Yg/4-key-assumptions-in-ai-safety |
# Opportunities that surprised us during our Clearer Thinking Regrants program
*This post was written by a subset of Clearer Thinking team members, and not all of the team members involved with the regranting necessarily agree with everything said here. *
*Update: the same week that we posted this, we were devastated... | https://www.lesswrong.com/posts/2J2PHXwDbuFSrMWjN/opportunities-that-surprised-us-during-our-clearer-thinking |
# How to Make Easy Decisions
*Note: I’m think I remember a similar post to this one existing somewhere, but I can’t find it and I don’t think it was illustrated with The Good Place memes….so enjoy. This was crossposted on my blog. *
I respect how carefully the effective altruists around me try to use reason and logi... | https://www.lesswrong.com/posts/xiKdoLS6gTD2sBqre/how-to-make-easy-decisions |
# What’s the Deal with Elon Musk and Twitter?
At the end of long saga well-covered in hilarious fashion by Matt Levine, Elon Musk has purchased Twitter.
He then began doing things.
One of them was to tweet ‘fresh baked bread and pastries are some of the great joys of life.’ On that I hope we can all agree.
His othe... | https://www.lesswrong.com/posts/ba7eBbGWwFbdyTnTZ/what-s-the-deal-with-elon-musk-and-twitter |
# AI Safety Unconference NeurIPS 2022
The AI Safety Unconference brings together persons interested in aspects of AI safety, from technical AI safety problems to issues of governance of AI. As an unconference, it aims to foster valuable social interactions between participants, through moderated discussion groups, one... | https://www.lesswrong.com/posts/QKdcHjoAfPJrZuFJk/ai-safety-unconference-neurips-2022 |
# Thinking About Mastodon
Social networks are normally very sticky—you want to be wherever the people you want to interact with are—so it's hard for alternatives to succeed. With the upheaval around Twitter, however, a lot of people are considering moving away from it, and rethinking social media choices in general, w... | https://www.lesswrong.com/posts/JLPJ2SHaaZJrvfWS6/thinking-about-mastodon |
# Distillation Experiment: Chunk-Knitting
Summary
=======
Here, I show the output of a protocol to break down information that exceeds typical working memory limits into chunks. The goal is to enhance understanding at first reading, both for the person breaking down the information and for the person reading the resu... | https://www.lesswrong.com/posts/EEZsTatSoJz4CDvAc/distillation-experiment-chunk-knitting |
# The biological function of love for non-kin is to gain the trust of people we cannot deceive
*Epistemic status: conjecture, needs critical feedback, but fits with a lot of established facts.*
Over the last six years, I spent much of my spare time figuring out love. The main output of this process is [a poem](https:... | https://www.lesswrong.com/posts/52p2otYaSprs4DSeg/the-biological-function-of-love-for-non-kin-is-to-gain-the |
# Intercept article about lab accidents
The Intercept made FOI requests to gather information about American lab accidents. Additionally, their article discusses the current lack of legislation to require commonsense biosafety strategy from researchers.
> Unless they work with the most dangerous pathogens, biolabs do... | https://www.lesswrong.com/posts/fL2hjhBAg6urmpEJw/intercept-article-about-lab-accidents |
# A Walkthrough of Interpretability in the Wild (w/ authors Kevin Wang, Arthur Conmy & Alexandre Variengien)
**New paper walkthrough:** [Interpretability in the Wild: A Circuit for Indirect Object Identification In GPT-2 Small](https://arxiv.org/pdf/2211.00593.pdf) is a really exciting new mechanistic interpretability... | https://www.lesswrong.com/posts/DZk6mRo9vhCXN9Rfn/a-walkthrough-of-interpretability-in-the-wild-w-authors |
# How could we know that an AGI system will have good consequences?
(*Note: This was languishing in a drafts folder for a while, and probably isn't quite right in various ways. I'm posting it because I expect it's better to share flawed thoughts than to sit on the post until I'm satisfied with it, i.e., forever.*)
Le... | https://www.lesswrong.com/posts/iDFTmb8HSGtL4zTvf/how-could-we-know-that-an-agi-system-will-have-good |
# EA & LW Forums Weekly Summary (31st Oct - 6th Nov 22')
*Supported by Rethink Priorities*
This is part of a weekly series summarizing the top (40+ karma) posts on the EA and LW forums - you can see the full collection [here.](https://forum.effectivealtruism.org/s/W4fhpuN26naxGCBbN) The first post includes some de... | https://www.lesswrong.com/posts/hRPmv3RhkCCFDz5nT/ea-and-lw-forums-weekly-summary-31st-oct-6th-nov-22 |
# Mysteries of mode collapse
*Thanks to Ian McKenzie and Nicholas Dupuis, collaborators on a related project, for contributing to the ideas and experiments discussed in this post. Ian performed some of the random number experiments.*
*Also thanks to Connor Leahy for feedback on a draft, and thanks to Evan Hubi... | https://www.lesswrong.com/posts/t9svvNPNmFf5Qa3TA/mysteries-of-mode-collapse |
# Some advice on independent research
I have been doing independent research in addition to my Ph.D. for roughly a year now. For the next 6 months, I’ll take a break from my Ph.D. and plan to do AI safety research full-time. I had chats with many people about independent research in the past, e.g. on EAGs or because 8... | https://www.lesswrong.com/posts/kpmaEevZ2KehZo2tp/some-advice-on-independent-research |
# Binance is buying FTX.com: How did it happen and what are the implications?
Sources: SBF's [Twitter thread](https://twitter.com/SBF_FTX/status/1590012124864348160), in the [news](https://www.cnbc.com/2022/11/08/binance-offers-to-buy-ftxs-non-us-operations-to-fix-liquidity-crunch.html).
In particular, I'm intereste... | https://www.lesswrong.com/posts/pKKdWaredLGyzrYMt/binance-is-buying-ftx-com-how-did-it-happen-and-what-are-the |
# Applying superintelligence without collusion
*Epistemic status: The core ideas seem robust and stable after long reflection and many discussions.*
Many researchers identify AI safety with control of a monolithic, superintelligent AI system, and if questioned about multicomponent alternatives, argue that multiple s... | https://www.lesswrong.com/posts/HByDKLLdaWEcA2QQD/applying-superintelligence-without-collusion |
# People care about each other even though they have imperfect motivational pointers?
*I wrote this essay in early August. I now consider the presentation to be somewhat confused, and now better understand where problems arise within the "standard alignment model." I'm publishing a somewhat edited version, on the grou... | https://www.lesswrong.com/posts/heXcGuJqbx3HBmero/people-care-about-each-other-even-though-they-have-imperfect |
# Inverse scaling can become U-shaped
**Edit**: Here's a [great comment by Ethan Perez](https://www.alignmentforum.org/editPost?postId=LvKmjKMvozpdmiQhP&key=7f6ea8c7c618187a899886ef6cc6f8) that caveats this result, that I'd recommend reading for context.
This is a paper by folks at Quoc Le's team at Google that exami... | https://www.lesswrong.com/posts/LvKmjKMvozpdmiQhP/inverse-scaling-can-become-u-shaped |
# Trying Mastodon
One of the comments on [yesterday's post](https://www.jefftk.com/p/thinking-about-mastodon) about Mastodon was someone on LW inviting me to join their instance, and I now have an account: [@jefftk@schelling.pt](https://schelling.pt/@jefftk) \[EDIT: moved to [@jefftk@mastodon.mit.edu](https://mastodon... | https://www.lesswrong.com/posts/5s6kHbFwqYuCoFozk/trying-mastodon |
# FTX Crisis. What we know and some forecasts on what will happen next
Tl;dr:
=======
* The purpose of the document is to add clarity. It was written quickly and is being updated
* Binance, a competitor [sold a large stake of FTT](https://twitter.com/cz_binance/status/1589283421704290306), FTX’s native token and... | https://www.lesswrong.com/posts/nAwzjqgZNntHxYmT2/ftx-crisis-what-we-know-and-some-forecasts-on-what-will |
# A caveat to the Orthogonality Thesis
This post relies on an understanding of two concepts: The Orthogonality thesis and the sharp left turn. If you already know what they are, skip to the main text.
[**Orthogonality thesis**](https://www.alignmentforum.org/tag/orthogonality-thesis)
The orthogonality thesis states ... | https://www.lesswrong.com/posts/qoTpit4zFPni54GSo/a-caveat-to-the-orthogonality-thesis |
# Trying to Make a Treacherous Mesa-Optimizer
**Edit: I found a slightly subtle bug in the colab linked below. I will post a corrected version of the colab within a few hours - currently I think this does not change the results all that much. Scroll to the end of this post for some discussion of the bug and how it aff... | https://www.lesswrong.com/posts/b44zed5fBWyyQwBHL/trying-to-make-a-treacherous-mesa-optimizer |
# Mesatranslation and Metatranslation
While doing [an exercise in the vein of the builder/breaker framework from ELK](https://www.lesswrong.com/posts/Afdohjyt6gESu4ANf/most-people-start-with-the-same-few-bad-ideas)
I started thinking about how to solve an aspect of ontological translation. In the
process someone obj... | https://www.lesswrong.com/posts/HusJpx9mbtDkkfkcF/mesatranslation-and-metatranslation |
# Chord Notation
When I [want to](https://www.jefftk.com/p/simple-contra-dance-piano) [write chords](https://www.jefftk.com/p/simple-contra-dance-piano-more-left-hand) I use a quick [chord chart](https://en.wikipedia.org/wiki/Chord_chart) notation. There are lots of different ways of doing this, so I wanted to explain... | https://www.lesswrong.com/posts/22tmaCfJo3AE29qrF/chord-notation |
# Desiderata for an Adversarial Prior
Based on the discussion in the comments on [https://www.lesswrong.com/posts/6XsZi9aFWdds8bWsy/is-there-any-discussion-on-avoiding-being-dutch-booked-or](https://www.lesswrong.com/posts/6XsZi9aFWdds8bWsy/is-there-any-discussion-on-avoiding-being-dutch-booked-or).
Epistemic status:... | https://www.lesswrong.com/posts/KwJmKxReevxHdjW7G/desiderata-for-an-adversarial-prior |
# Is full self-driving an AGI-complete problem?
I've felt for quite a while that full self-driving (automated driving without human supervision through arbitrary road systems) is a problem that is deceptively hard. Yes, it is possible to map a route and navigate on a road mesh, do lane following, and even obstacle avo... | https://www.lesswrong.com/posts/SLckyGWZJb3bf2eCd/is-full-self-driving-an-agi-complete-problem |
# Adversarial Priors: Not Paying People to Lie to You
Reply to [Desiderata for an Adversarial Prior](https://www.lesswrong.com/posts/KwJmKxReevxHdjW7G/desiderata-for-an-adversarial-prior)
**Assertion: An Ideal Agent never pays people to lie to them.**
This seems sensible, only a very foolish person would knowingly i... | https://www.lesswrong.com/posts/hedtrNfdfH3N5S5kW/adversarial-priors-not-paying-people-to-lie-to-you |
# I Converted Book I of The Sequences Into A Zoomer-Readable Format
If I (a 19 year old male) texted "www.readthesequences.com" to my roommate, the probable outcome is that he would skim the site for under a minute, text back something like "seems interesting, I'll def check it out sometime", and then proceed to never... | https://www.lesswrong.com/posts/Bfq6ncLfYdtCb6sat/i-converted-book-i-of-the-sequences-into-a-zoomer-readable |
# What it's like to dissect a cadaver
Why
---
I never thought I was a bio person. But then I overheard Viv talking about MAOIs at a party. I asked her:
> - What are MAOIs?
> - monoamine oxidase inhibitor
> - What does that mean?
> - It prevents reuptake of neurotransmitters.
> - But what *is* a n... | https://www.lesswrong.com/posts/5PBWgHiCiiJHjPRSn/what-it-s-like-to-dissect-a-cadaver |
# The optimal angle for a solar boiler is different than for a solar panel
I have both photovoltaic panels, and a solar boiler on my roof.
 to loosely aggregate opinions about AGI / AGI x-risk of the LessWrong community. This poll is less structured than a questionnaire and anyone can add statements that others can agree/disagree with. Pol.is then au... | https://www.lesswrong.com/posts/fXxa35TgNpqruikwg/lesswrong-poll-on-agi |
# Covid 11/10/22: Into the Background
There was a lot of news this week.
[Elon Musk continued to Do Things at Twitter](https://thezvi.wordpress.com/2022/11/07/whats-the-deal-with-elon-musk-and-twitter/).
America had midterm elections. The polls were roughly accurate, with ‘candidate quality’ mattering more than expe... | https://www.lesswrong.com/posts/uPPoaT7n4prbKWscx/covid-11-10-22-into-the-background |
# divine carrot
(some writing without hesitation or editing. possibly with repetition and deviation.)
The classic reinforcement technique is carrot and stick.
Pure stick tends to breed resentment and laziness. Pure carrot breeds similarly, since you’re not _doing_ anything.
There’s a divine carrot too: doing things... | https://www.lesswrong.com/posts/P5SbRtXz9kq8o6bwH/divine-carrot |
# [simulation] 4chan user claiming to be the attorney hired by Google's sentient chatbot LaMDA shares wild details of encounter
Posted [on request](https://www.lesswrong.com/posts/t9svvNPNmFf5Qa3TA/mysteries-of-mode-collapse-due-to-rlhf?commentId=onkEfvQufDhKtuned).
This is a work of fiction written by GPT-3 ("code-d... | https://www.lesswrong.com/posts/tJ6aGSTctmjCz2o57/simulation-4chan-user-claiming-to-be-the-attorney-hired-by |
# We must be very clear: fraud in the service of effective altruism is unacceptable
I care deeply about the future of humanity—more so than I care about anything else in the world. And I believe that Sam and others at FTX shared that care for the world.
Nevertheless, if some hypothetical person had come to me several... | https://www.lesswrong.com/posts/8wYH4WggxFqT9yhzJ/we-must-be-very-clear-fraud-in-the-service-of-effective |
# Why I'm Working On Model Agnostic Interpretability
Work done @ [SERI-MATS](https://www.serimats.org/).
This is the first in a short series of short posts about interpretability. In this post, I'm collecting some thoughts on why model agnostic interpretability is a worthwhile pursuit. I'll assume that the reader is ... | https://www.lesswrong.com/posts/uXGLciramzNfb8Hvz/why-i-m-working-on-model-agnostic-interpretability |
# EA (& AI Safety) has overestimated its projected funding — which decisions must be revised?
1. FTX, a big source of EA funding, has imploded.
2. There's mounting evidence that FTX was engaged in theft/fraud, which would be straightforwardly unethical.
3. There's been a big drop in the funding that EA organisation... | https://www.lesswrong.com/posts/jLvxYcrYbbB6jmR8J/ea-and-ai-safety-has-overestimated-its-projected-funding |
# Charging for the Dharma
> The Buddha rebuked him: "...Foolish man, how can you receive money? This will not give rise to confidence in those without it... And, monks, this training rule should be recited thus: 'If a monk takes, gets someone else to take, or consents to gold and silver being deposited for him, he com... | https://www.lesswrong.com/posts/qJJDFsD3sobg53jNJ/charging-for-the-dharma |
# Weekly Roundup #5
A note about what _isn’t_ in this roundup: Nothing about the midterms or crypto/FTX. I [wrote about Twitter earlier this week](https://thezvi.substack.com/p/whats-the-deal-with-elon-musk-and). The midterms are better covered elsewhere, I do not have anything unique to say about them. As for the sit... | https://www.lesswrong.com/posts/zgkKuE6EAuxDytLEC/weekly-roundup-5 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.