id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
2f44d1b4-ce45-4bf4-bb8b-84c3384b073d
trentmkelly/LessWrong-43k
LessWrong
Natural Categories Update One concern in the AI-Alignment problem is that neural networks are "alien minds".  Namely the representations that they learn of the world are too weird/different to allow effect communication of human goals and ideas. For example, EY writes > there is no known way to use the paradigm of l...
936af9ce-3aa0-48a8-ae2b-00f5c2c1ee6e
trentmkelly/LessWrong-43k
LessWrong
How Death Feels [Epistemic Status: rambly and exploratory, hoping for some serious discussion] ---------------------------------------- It's 2030, the AGI takeoff ended badly and humanity is wiped out by a misaligned AGI. You die. The end. The end? The above is a cliché many of us have in mind when discussing scena...
20ca154d-c5dc-4ded-a728-684b886c1fc3
trentmkelly/LessWrong-43k
LessWrong
The Worst Form Of Government (Except For Everything Else We've Tried) Churchill famously called democracy “the worst form of Government except for all those other forms that have been tried from time to time” - referring presumably to the relative success of his native Britain, the US, and more generally Western Europ...
35c7873f-fdf7-4639-97e6-06b559ff455f
trentmkelly/LessWrong-43k
LessWrong
Syllable Status: fun to write, very derivative, ambivalent about content. This post responds to Why and How to Name Things. In the evolution of language, new words are constantly produced, bubbling in and out of primordial concept-space like quantum foam. Most of them are awkward, unevocative, and noncentral, fallin...
20cf2498-11a6-43aa-9eda-b4647ec8637e
trentmkelly/LessWrong-43k
LessWrong
Machine Learning Consent For years, researchers have trained machine learning systems on whatever data they could find. People mostly haven't cared about this or paid attention, I think because the systems hadn't been very good. Recently, however, some very impressive systems have come out, including ones that answer ...
cc60338f-0cbd-40ee-ab6e-c980e5a69cd6
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Resources for AI Alignment Cartography **I want to make an *actionable* map of AI alignment.** After years of reading papers, blog posts, online exchanges, books, and occasionally hidden documents about AI alignment and AI risk, and having extremely interesting conversations about it, most arguments I encounter now f...
d857cc0f-7eb1-4c4d-a74d-cf8ef15b578a
trentmkelly/LessWrong-43k
LessWrong
Labor Participation is an Alignment Risk TLDR: This paper explores and outlines why AI-related reductions in Labor participation (e.g., real unemployment) is a significant and high-priority alignment risk that the alignment community should more seriously consider as a potential underlying systemic risk factor. Constr...
0e7c2fa6-b8c1-48d1-91bd-f2043f9b4bd0
trentmkelly/LessWrong-43k
LessWrong
What is Morality? The PDF version can be read here. The audio version can be listened to here: Morality has the following components: * Collective values. * The individual internalization of collective values. * The assumption that collective values are objective, and thus “moral”. * A folk theory of morality....
75d48405-625d-4d49-af50-128b02153a91
trentmkelly/LessWrong-43k
LessWrong
Earning to Give vs. Altruistic Career Choice Revisited A commonly voiced sentiment in the effective altruist community is that the best way to do the most good is generally to make as much money as possible, with a view toward donating to the most cost-effective charities. This is often referred to as “earning to give...
134bce03-235f-4f12-b033-e6b4a3856a11
StampyAI/alignment-research-dataset/blogs
Blogs
explaining "." explaining "." -------------- i occasionally send people the following message, on text chat platforms: > . > > i've been asked what it means often enough that at this point i'm making a short blogpost. "." is the empty message. it conveys whatever it means to say nothing, but explicitely — w...
d298a4f1-78ea-41be-90f6-4b13bd7e056e
trentmkelly/LessWrong-43k
LessWrong
The Correct Use of Analogy In response to: Failure by Analogy, Surface Analogies and Deep Causes Analogy gets a bad rap around here, and not without reason. The kinds of argument from analogy condemned in the above links fully deserve the condemnation they get. Still, I think it's too easy to read them and walk away ...
daf32422-4fe9-4c7d-9faf-6059b157b3ff
StampyAI/alignment-research-dataset/blogs
Blogs
the Belief In Society compass the Belief In Society compass -----------------------------
5f2ba278-4be7-434d-9417-b6bd295f6428
trentmkelly/LessWrong-43k
LessWrong
Compression Moves for Prediction Imagine that you want to predict the behavior of some system -- the weather, a computer program, a real-world language... However you go about it, I bet you will do some compression: you'll find simple and minimal and concrete ways to think about the vast complexity of your system that...
d1c923b9-b6f8-47ba-95af-fe6f0db61897
StampyAI/alignment-research-dataset/arxiv
Arxiv
Quantifying the pathways to life using assembly spaces 1 Quantifying the pathways to life using assembly spaces Stuart M. Marshall,1 Douglas G. Moore,2 Alastair R. G. Murray,1 Sara I. Walker,2,3* and Leroy Cronin1* 1 School of Chemistry, University of Glasgow, Glasgow, G12 8QQ, UK. 2 BEYOND Center for Fundam...
e76fdf05-68ef-48fa-b722-7983deeed116
StampyAI/alignment-research-dataset/special_docs
Other
QNRs: Toward Language for Intelligent Machines >FHI TECHNICAL REPORT < QNRs Toward Language for Intelligent Machines K. Eric Drexler Senior Research Fellow eric.drexler@oxfordmartin.ox.ac.uk Technical Report #2021-3 Cite as: Drexler, K.E. (2021): “QNRs: Toward Language for Intelligent Machines”, Technical Report #2021...
4b7c233e-e7b2-437a-8f31-188fe3ae2e20
trentmkelly/LessWrong-43k
LessWrong
Alignment as Function Fitting In this post, I spell out some fairly basic ideas, that seem important to me. Maybe they are so basic that everyone finds them obvious, but in the spirit of this piece, I thought I would write them up anyway. tl;dr I start off by looking at using linear regression to model data. I examin...
dab412ba-5fda-4e4c-ae50-b0ede39f7dea
trentmkelly/LessWrong-43k
LessWrong
Interview with Aella, Part I Cross-posted from Putanumonit. Things that happen offline notice: if you’re reading this I strongly predict that you will enjoy attending a Slate Star Codex meetup. I will be at the New York meetup on Saturday 9/21, come say hi! This is part 1 of the interview, talking about Aella’s back...
02e6ccd8-c4a3-4d34-9496-b3b632f7e0e6
StampyAI/alignment-research-dataset/blogs
Blogs
Imagining yourself as a digital person (two sketches) Today’s world Transformative AI Digital people World of Misaligned AI World run by Something else or or Stable, galaxy-wide civilization > > This is a short, casual piece about [digital people](https://www.cold-takes.com/how-digital-...
7e286c41-fa77-4cc8-b06b-8d4a5b97369c
trentmkelly/LessWrong-43k
LessWrong
[ASoT] Simulators show us behavioural properties by default I’m trying out making a few posts with less polish and smaller scope, to iterate more quickly on my thoughts and write about some interesting ideas in isolation before having fully figured them out. Expect middling confidence in any conclusions drawn, and occ...
98b843c4-9594-40b0-a4aa-60ee52ae70f4
trentmkelly/LessWrong-43k
LessWrong
Meetup : SF Meetup: Projects Discussion article for the meetup : SF Meetup: Projects WHEN: 12 December 2016 06:15:07PM (-0800) WHERE: 1769 15th St., SF We’ll be meeting to work on projects! Near the beginning, we’ll go around and talk about what we’ll be working on, then do a couple of pomodoros quietly. At some p...
62517e02-e221-4035-8f14-640f4c5a82ac
trentmkelly/LessWrong-43k
LessWrong
Vision of a positive Singularity Introduction Many people feel significant anxiety about how AI and superintelligence will play out, regarding both the ultimate outcome and the intermediate stages. There is the sense that some kind of loss is inevitable as humanity becomes more powerful in this way. There is also the...
67b40faf-53d3-4302-b366-fe5a4b460d76
trentmkelly/LessWrong-43k
LessWrong
Exercise in rationality: popular quotes, revisited A friend recently shared an image of Lincoln with the quote, "Better to remain silent and be thought a fool than speak and remove all doubt."   Correcting that idea, I replied with the following: "Speak! Reveal your foolishness, and open yourself so that others may ...
ba6103a1-0ebb-4d37-9a8f-ce910f5ac9cd
trentmkelly/LessWrong-43k
LessWrong
Rationality Quotes 27 "Believing this statement will make you happier."         -- Ryan Lortie "Make changes based on your strongest opportunities, not your most convenient ones."         -- MegaTokyo "The mind is a cruel, lying, unreliable bastard that can't be trusted with even an ounce of responsibility.  If you ...
49684391-e882-4937-8e04-26af40e7d1bd
StampyAI/alignment-research-dataset/arbital
Arbital
List A list is an ordered collection of objects, such as `[1, 2, 3](https://arbital.com/p/0,)` or `["blue", 0, "shoe"](https://arbital.com/p/"red",)`. Lists are generally denoted by square brackets, with elements separated by commas. Lists allow repetition (unlike [sets](https://arbital.com/p/3jz)) and are ordered (un...
4041dc52-3d28-459f-b9dc-49eb61ddd462
trentmkelly/LessWrong-43k
LessWrong
United We Align: Harnessing Collective Human Intelligence for AI Alignment Progress Summary: Collective Human Intelligence (CHI) represents both the current height of general intelligence and a model of alignment among intelligent agents. However, CHI's efficiency and inner workings remain underexplored. This research...
e47edf46-3610-45c4-a83c-6277c5af01df
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Protectionism will Slow the Deployment of AI I believe I’m more optimistic than the average LWer that regulation will be passed that slows down AI capabilities research. The current capabilities of language models threaten the interests of the politically salient professional managerial middle class, which will cause ...
b63c1cf1-1204-42ea-99a1-e468dce7f866
trentmkelly/LessWrong-43k
LessWrong
Where are the Benefits from Conversation? (epistemic status: fairly speculative) Recently, I've been thinking about where the main benefits of conversation lie, and I've come to some interesting early conclusions. Most of the benefits of private conversations come from effects they have on the immediate participants...
84df7987-557d-4bc5-a32c-3d2daed8943f
trentmkelly/LessWrong-43k
LessWrong
Call for new SIAI Visiting Fellows, on a rolling basis Last summer, 15 Less Wrongers, under the auspices of SIAI, gathered in a big house in Santa Clara (in the SF bay area), with whiteboards, existential risk-reducing projects, and the ambition to learn and do. Now, the new and better version has arrived.  We’re tak...
a2447905-8fc1-4c3e-8e57-5421b2b311c9
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post3111 In " Risks from Learned Optimization ," we talked about a variety of different forms of pseudo-alignment—that is, ways in which a trained model's objective (its mesa-objective ) can be misaligned off-distribution with the loss function it was trained under (the base objective ). In particular, we distingu...
a2e63511-9fbb-4392-a117-70a3f3d25f76
trentmkelly/LessWrong-43k
LessWrong
User-Controlled Algorithmic Feeds I've been playing with Mastodon more, and the main way you see what other people have posted is by reading your feed. This consists of everything any of the people you follow has posted or boosted. If you're always going to read everything that's come out since your last visit, which ...
49905707-4a44-48a6-89e4-d060673791c8
trentmkelly/LessWrong-43k
LessWrong
Uninformed Elevation of Trust I don't know if there is a standard name for this phenomenon (other than the related Gell-Mann amnesia effect, e.g. "The Sequences are great, except in my area of expertise, where they are terrible."). Here is the gist: we trust the data as much as we trust the source, regardless of how ...
9671bcf4-4fec-40a9-938c-fd7c585da9c8
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] On Being Decoherent Today's post, On Being Decoherent was originally published on 27 April 2008. A summary (taken from the LW wiki):   > When a sensor measures a particle whose amplitude distribution stretches over space - perhaps seeing if the particle is to the left or right of some dividing line - the...
681f3fed-4f42-4c21-bfd0-50617e9f6c55
trentmkelly/LessWrong-43k
LessWrong
Weekly Wiki Workshop and suggested articles Now 12 days old, LWiki is still growing. Most of the articles are stubs, but progress is being made. Eliezer confirmed that an official wiki hosted on LW is eventually coming, so be careful about linking to the wiki, but don't let that deter you from adding content.  Standa...
0f651f01-2b8f-451c-840e-a04290d53717
trentmkelly/LessWrong-43k
LessWrong
The Bat and Ball Problem Revisited Cross posted from my personal blog. In this post, I'm going to assume you've come across the Cognitive Reflection Test before and know the answers. If you haven't, it's only three quick questions, go and do it now. One of the striking early examples in Kahneman's Thinking, Fast and...
00c789af-6537-4c8d-b730-ba03ec52cca8
trentmkelly/LessWrong-43k
LessWrong
How can you use music to boost learning? I often find that I am able to appreciate the beauty of a subject more while listening to music (especially instrumental music). Hearing the notes while I think about the topic helps creates a lot of subconscious connections with the material, solidifying what I am learning as ...
0949f116-fbf6-4d96-9023-e6e2bb447c05
trentmkelly/LessWrong-43k
LessWrong
Observations on Teaching for Four Weeks I just finished a program where I taught two classes of high school seniors, two classes a day for four weeks, as part of my grad program.  This experience was a lot of fun and it was rewarding, but it was really surprising, and even if only in small ways prompted me to update ...
4715c8b1-05e1-4796-b6d2-950f6076696d
trentmkelly/LessWrong-43k
LessWrong
Can LLMs Simulate Internal Evaluation? A Case Study in Self-Generated Recommendations Author: Yap Chan Chen Independent Researcher, Malaysia chanchen83@hotmail.com Abstract: This post presents an exploratory study into how large language models (LLMs) respond to prompts that invite them to simulate evaluative judgmen...
15446e8a-c2f6-4e5b-8f17-9404231b6736
StampyAI/alignment-research-dataset/blogs
Blogs
unoptimal superintelligence loses unoptimal superintelligence loses --------------------------------- (edit: [maybe it doesn't](unoptimal-superint-doesnt-lose.html)) what if a phenomena is powerful enough to kill everyone, but not smart enough to be optimal at reasoning? (such as a grey goo event, or a "dumb" supe...
4ed37d95-3884-441d-ba90-8922490415f5
trentmkelly/LessWrong-43k
LessWrong
20 minutes of work as an artist in one future I am an artist. ”Eleven evil wizard schoolgirls in an archduke's library, dressed in red and black Asmodean schoolgirl uniforms, perched on armchairs and sofas”[1] Sigh, at least it’s not more catgirls. I don’t even draw them well. I stretched briefly before starting...
f8d5edfb-18d1-40a0-8ead-bf930fc57df9
trentmkelly/LessWrong-43k
LessWrong
How to teach things well (This is a post on my thoughts on good teaching techniques from a daily blogging project, that I thought might be of interest to LessWrong readers) Introduction This is a blog post on how to teach things well. I’ll mostly be focusing on forms of teaching that involve preparation and structur...
2305b4b0-e34e-44e1-affe-ff6edba2156b
trentmkelly/LessWrong-43k
LessWrong
A simple-minded theory of "observer fluid" Eliezer has proposed a puzzle: > So here's a simple algorithm for winning the lottery: > > Buy a ticket.  Suspend your computer program just before the lottery drawing - which should of course be a quantum lottery, so that every ticket wins somewhere.  Program your computat...
09d03783-9e9f-47d9-87bf-16d34bc860be
StampyAI/alignment-research-dataset/agentmodels
Tutorial: Modeling Agents with Probabilistic Programs
Modeling Agents with Probabilistic Programs --- layout: chapter title: Reasoning about agents description: Overview of Inverse Reinforcement Learning. Inferring utilities and beliefs from choices in Gridworld and Bandits. is_section: true --- ## Introduction The previous chapters have shown how to compute optimal ac...
fff79916-6faf-43e3-873d-1000a0873db1
trentmkelly/LessWrong-43k
LessWrong
Once we cure aging, what happens to prisoners serving life sentences? Biogerontologist Aubrey de Grey claimed that the first person who will live to 1,000 years old has already been born. He believes this unprecedented expansion in lifespan will occur with Longevity Escape Velocity (LEV). LEV is defined by technologi...
8b3ecba9-b1e9-43d6-af08-e5ec1ef5076c
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
PhD student and postdoc positions philosophy of AI in Erlangen (Germany) The Centre for Philosophy and AI Research in Erlangen, Germany, directed by Prof. Vincent Müller, is looking for four (funded) PhD students, one postdoctoral researcher and one director of research. This is a great chance for philosophers, or asp...
8b4013fb-ed69-4a94-840b-4d860ec63316
trentmkelly/LessWrong-43k
LessWrong
"Building a House" Review I was recently reading Byron Barton's 1981 book, Building a House. While it claims to be an end-to-end overview of the process of modern (for the time) home construction, there are enough errors in the illustrations that I wouldn't recommend it as a basic text. For example, here's how they s...
9719c951-92e8-4973-9f60-709741010b91
trentmkelly/LessWrong-43k
LessWrong
"But It Doesn't Matter" If you ever find yourself saying, "Even if Hypothesis H is true, it doesn't have any decision-relevant implications," you are rationalizing! The fact that H is interesting enough for you to be considering the question at all (it's not some arbitrary trivium like the 1923th binary digit of π, or...
b3f615ef-af11-41b7-9613-0080aaaf2169
trentmkelly/LessWrong-43k
LessWrong
Eight Short Studies On Excuses The Clumsy Game-Player You and a partner are playing an Iterated Prisoner's Dilemma. Both of you have publicly pre-committed to the tit-for-tat strategy. By iteration 5, you're going happily along, raking up the bonuses of cooperation, when your partner unexpectedly presses the "defect"...
22fbe90a-a72b-460b-ba64-8b6ef4130618
trentmkelly/LessWrong-43k
LessWrong
Rationality Quotes July 2016 Another month, another rationality quotes thread. The rules are: * Provide sufficient information (URL, title, date, page number, etc.) to enable a reader to find the place where you read the quote, or its original source if available. Do not quote with only a name. * Post all quotes se...
8b1a2cdd-e1c0-45e9-9dfe-71bd5592729f
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"(Cross-posted from the EA Forum)This post was written by Buck and Claire Zabel but it’s written in Buck’s voice, and “I” here refers to Buck, because it’s about grantmaking that he might do. (Claire contributed in her personal capacity, not as an Open Phil grantmaker). In addition to accepting applications for EA grou...
4ce9cd44-787d-42b1-bb4f-c89dcbd3af6a
trentmkelly/LessWrong-43k
LessWrong
Conjecture internal survey: AGI timelines and probability of human extinction from advanced AI We put together a survey to study the opinions of timelines and probability of human extinction of the employees at Conjecture. The questions were based on previous public surveys and prediction markets, to ensure that the r...
d74338e2-de37-422b-add9-845ee167fe10
trentmkelly/LessWrong-43k
LessWrong
How to impress students with recent advances in ML? I am preparing resources for an equivalent of the mlab camp, organized by EffiSciences,  that will be held in France in a month to train students to become prosaic alignment researchers. 3/4 of the participants are already aligned with existential risks related to A...
163956be-f0f5-4c03-b740-60cb1818b124
trentmkelly/LessWrong-43k
LessWrong
Accidental Terraforming On the subject of climate change - specifically anthropogenic (human-caused) climate change - much ink has been spilled. Or keys typed, I suppose, since the internet doesn’t actually use ink. Regardless, the Great War over acknowledging the existence of Climate Change seems to be over; enough ...
533b6768-8ebe-4081-9efd-e789405d4f51
trentmkelly/LessWrong-43k
LessWrong
Build a Metaculus Forecasting Bot in 30 Minutes: A Practical Guide We've created a video tutorial showing how to build a forecasting bot using our template in about 30 minutes. The template bot consistently ranks in the top 10 of Metaculus's AI Forecasting Benchmark Series using a basic GPT-4/AskNews implementation. ...
b8b11a3d-5cd7-429a-bf61-d7ba38b21ec8
trentmkelly/LessWrong-43k
LessWrong
Interpreting Modular Addition in MLPs Summary In this post, we investigate one of Neel Nanda's 200 Concrete Open Problems in Mechanistic Interpretability, namely problem A3.3: Interpret a 2L MLP (one hidden layer) trained to do modular addition. The network seems to learn the following function: logit(c)∝∑Ni=0relu(u...
77f792da-4f9f-4f8c-b338-f82c90ca52d9
trentmkelly/LessWrong-43k
LessWrong
The Inner Ring by C. S. Lewis Note: In @Nathan Young's words "It seems like great essays should go here and be fed through the standard LessWrong algorithm. There is possibly a copyright issue here, but we aren't making any money off it either."  What follows is a full copy of the C. S. Lewis essay "The Inner Ring" t...
ae368684-1726-4c28-9abe-82313c980728
trentmkelly/LessWrong-43k
LessWrong
What to do after inventing AGI? Imagine a research team makes a breakthrough when experimenting with neural networks. They train a model with only 1 billion parameters, and it displays general intelligence that exceeds GPT-3s. They conclude that a 100 billion parameter model, possible to train for a few million dollar...
af6a87d2-fb9f-4c2f-b13c-062fd5026750
trentmkelly/LessWrong-43k
LessWrong
Back to the Past to the Future   You are sitting in your bedroom, dorm, tent, sidewalk, wherever you are staying right now, reading these words, when a dark void appears in your peripheral vision. A figure walks out of the void. They look a bit different, but also familiar. Surprise! It’s you from the future! And you...
9b2563bc-b845-466e-b0d6-037aa3f5bbf0
trentmkelly/LessWrong-43k
LessWrong
Are there English-speaking meetups in Frankfurt/Munich/Zurich? I'll be in Stuttgart, Germany for a few months this summer for work. It's about equidistant from Frankfurt, Munich, and Zurich. Do any of these cities have regular meetups where English is spoken, at least in part? About me -- I run a software startup, wh...
705c8d0b-3e19-4e02-b2f8-169ebcf45881
trentmkelly/LessWrong-43k
LessWrong
Researchers believe they have found a way for artists to fight back against AI style capture The real details are here: What is Glaze? I wonder how robust this technique is to adversarial attacks. It claims to work at the pixel level, masking an image by distorting the pixels in ways invisible to the human eye but wh...
8fbe4703-5383-4b9c-be6f-9025ab8791d5
trentmkelly/LessWrong-43k
LessWrong
How Do You Interpret the Goal of LessWrong and Its Community? I lurk LessWrong and am grappling with a perceived misalignment between its stated goals—improving reasoning and decision-making—and the type of content often shared. I am not referring to content that I disagree with, or content that I think is poorly writ...
a873e7c2-63d3-48fb-8d35-fc17c2704161
trentmkelly/LessWrong-43k
LessWrong
Volunteers needed to work on LessWrong's public goods problem I want to figure out how to solve LessWrong's public goods problem. There are lots of projects that lots of people are moderately interested in but not interested enough to do themselves. For example: * Further contributions to the Science of Winning at L...
f9101a4e-0b16-4dd3-8179-6c3103a424a8
trentmkelly/LessWrong-43k
LessWrong
Bayes Slays Goodman's Grue This is a first stab at solving Goodman's famous grue problem. I haven't seen a post on LW about the grue paradox, and this surprised me since I had figured that if any arguments would be raised against Bayesian LW doctrine, it would be the grue problem. I haven't looked at many proposed sol...
0a71902c-97e2-408c-a002-a5266dd96602
trentmkelly/LessWrong-43k
LessWrong
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). In this post, I’ll explain how the GENIES benchmark relates to a b...
6f2ddd66-397d-4a12-a9d6-d0181f9260aa
trentmkelly/LessWrong-43k
LessWrong
What is an "anti-Occamian prior"? I've seen references to "anti-Occamian priors" in the Sequences, where Eliezer was talking about how not all possible minds would agree with Occam's Razor. I'm not sure how such a prior could consistently exist. It seems like Occam's Razor just logically follows from the basic premis...
f30f2a88-8d40-4604-b609-a27e6530e43f
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"1. Money wants to be linear, but wants even more to be logarithmic.People sometimes talk as if risk-aversion (or risk-loving) is irrational in itself. It is true that VNM-rationality implies you just take expected values, and hence, don't penalize variance or any such thing. However, you are allowed to have a concave ...
b9761685-9fb0-43a8-9fd0-ea3abc5a7c12
trentmkelly/LessWrong-43k
LessWrong
Misnaming and Other Issues with OpenAI's “Human Level” Superintelligence Hierarchy Bloomberg reports that OpenAI internally has benchmarks for “Human-Level AI.” They have 5 levels, with the first being the achieved level of having intelligent conversation, to level 2, “[unassisted, PhD-level] Reasoners,” level 3, “Age...
8a9b06ef-3d7b-471c-8b11-83587fc1934e
trentmkelly/LessWrong-43k
LessWrong
The Center for Sustainable Nanotechnology Those concerned about existential risks may be interested to learn that, as of last September, the National Science Foundation is funding a Center for Sustainable Nanotechnology.  Though I haven't yet seen anywhere where they explicitly characterize nanotechnology as an existe...
941ec101-0904-433a-b44b-aa486c6562d9
trentmkelly/LessWrong-43k
LessWrong
Calling all MIRI supporters for unique May 6 giving opportunity! (Cross-posted from MIRI's blog. MIRI maintains Less Wrong, with generous help from Trike Apps, and much of the core content is written by salaried MIRI staff members.) Update: I'm liveblogging the fundraiser here. Read our strategy below, then give he...
77c0cbeb-cc74-47da-8eb3-a005b847a7c3
trentmkelly/LessWrong-43k
LessWrong
Meetup : Believing and Deciding - Melbourne Rationality Dojo, May 2017 Discussion article for the meetup : Believing and Deciding - Melbourne Rationality Dojo, May 2017 WHEN: 21 May 2017 03:30:00PM (+1000) WHERE: Henley Club Lets run though some of the rationality community's favourite lessons on believing and deci...
cadb9cfc-0db8-4cd9-bce0-ce09d9f17698
trentmkelly/LessWrong-43k
LessWrong
[Video] Why Not Just: Think of AGI Like a Corporation? (Robert Miles) Robert Miles has been creating AI-Alignment related videos for a while now, but I found this one particularly good. Here is the automatically generated Youtube transcript. Obviously it's not very good, but at least it makes the post searchable (In ...
a003e198-514a-4d7e-a757-708ed3f9efd4
trentmkelly/LessWrong-43k
LessWrong
Nash equilibriums can be arbitrarily bad Go hungry with Almost Free Lunches Consider the following game, called "Almost Free Lunches" (EDIT: this seems to be a variant of the traveller dilemma). You name any pound-and-pence amount between £0 and £1,000,000; your opponent does likewise. Then you will both get whichever...
712cb6b9-8ab6-49ba-a438-8ce8be153b7b
trentmkelly/LessWrong-43k
LessWrong
A simulation basilisk Important note: although I use the term "basilisk" to refer to this situation because of its similarities with the famous one, note that neither this situation nor the other famous "basilisk" are actually dangerous for humans to think about, they only affect superintelligences. Suppose that we a...
d5dee537-c4eb-47b9-b911-8086785b6a67
trentmkelly/LessWrong-43k
LessWrong
New edition of "Rationality: From AI to Zombies" MIRI is releasing a new edition of Rationality: From AI to Zombies, including the first set of R:AZ print books. As of this morning, print versions of Map and Territory (volume 1) and How to Actually Change Your Mind (volume 2) are now available on Amazon (1, 2), and we...
6b5c2747-176d-4a42-993e-c8f694c542c4
trentmkelly/LessWrong-43k
LessWrong
A map: Causal structure of a global catastrophe Pdf: http://immortality-roadmap.com/levelglobcat.pdf
b95a8f1d-9738-441b-9fd2-0e2b05e0ec87
trentmkelly/LessWrong-43k
LessWrong
Learn Three Things Every Day In the Game of Thrones series, there is an ongoing side plot in which a character is trained by a secretive organization to become an assassin. As part of her training, one of the senior assassins demands that she report to him three new things she has learnt every day. by making a natural...
dae0638b-271d-490c-8237-84d6d4df42ea
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Defining Myopia IID vs Myopia ============= In [a comment to Partial Agency](https://www.lesswrong.com/posts/4hdHto3uHejhY2F3Q/partial-agency#kkejCdMFEoD2T2874), Rohin summarized his understanding of the post. He used the iid assumption as a critical part of his story. Initially, I thought that this was a good descri...
35811e76-b977-4fa1-9782-33b03179ff8f
trentmkelly/LessWrong-43k
LessWrong
Speculation on Path-Dependance in Large Language Models. Epistemic Status: Highly Speculative. I spent less than a day thinking about this in particular, and though I have spent a few months studying large language models, I have never trained a language model. I am likely wrong about many things. I have not seen rese...
3cf2332c-f6f1-44c1-af7a-ece5c3f84d28
trentmkelly/LessWrong-43k
LessWrong
Hydroxychloroquine: the day science stopped This is a translation of (what I consider the most important bits) part (32%) of the original article "Hidroxicloroquina: o dia em que a ciência parou" from Luis Cláudio Correia, posted on Evidence Based Medicine Blog, 20th March 2020. Please correct mispellings, any English...
d0b3487b-64fc-49a5-abd0-8c247c392dcb
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Conditioning Predictive Models: Interactions with other approaches *This is the fifth of seven posts in the [Conditioning Predictive Models Sequence](https://www.alignmentforum.org/s/n3utvGrgC2SGi9xQX) based on the paper “[Conditioning Predictive Models: Risks and Strategies](https://arxiv.org/abs/2302.00805)” by Evan...
2b758d57-972d-4c81-af20-7a03794f41bd
trentmkelly/LessWrong-43k
LessWrong
Introducing EffiSciences’ AI Safety Unit  This post was written by Léo Dana, Charbel-Raphaël Ségerie, and Florent Berthet, with the help of  Siméon Campos, Quentin Didier, Jérémy Andréoletti, Anouk Hannot, Angélina Gentaz, and Tom David.     In this post, you will learn what were EffiSciences’ most successful fiel...
8ae48421-e788-430f-ad14-08ed038790ac
trentmkelly/LessWrong-43k
LessWrong
LW UI issue Not really sure where else I might post this, but there seems to be a UI issue on the site. When I hit the homepage of lesswrong.com while logged in I no longer see the user sidebar or the header links for Main and Discussion. This is kind of annoying because I have to click into an article first to get to...
ec04755c-3cfa-4351-a8bb-70489e5c59e9
trentmkelly/LessWrong-43k
LessWrong
Fixed Point Discussion ,, Warning: This post contains some important spoilers for Topological Fixed Point Exercises, Diagonalization Fixed Point Exercises, and Iteration Fixed Point Exercises. If you plan to even try the exercises, reading this post will significantly reduce the value you can get from doing them. ---...
c8a654a6-6f01-4444-9e6e-001cd39ea051
trentmkelly/LessWrong-43k
LessWrong
Babble challenge: 50 ways to escape a locked room Once again, it is time to become stronger.  This week’s challenge: You find yourself in a locked, empty room. You only have the clothes on your body and a phone in your pocket. You have enough energy to not need food or water for 10 years. Your phone has enough batter...
eab9b874-1723-4d87-912a-f0b509367751
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
Bad AI Predictions: Bard Upgrade, 2 Years to AI Auto-Money, OpenAI Investigation and more people are starting to get used to the Breakneck pace of AI so I wanted to show that developments in just the last few days fly in the face of what was predicted only a couple of years ago starting with the upgrades The Bard then...
49154d2d-171e-4993-a123-a2f49bdb832f
trentmkelly/LessWrong-43k
LessWrong
Games for factoring out variables All the methods proposed for factoring out B (for having the AI maximise a certain value while 'ignoring' it's impact via B) can be put on the same general footing. For some set A, define a function Q on A×A with Q(a,a′)≥0 and ∑a,a′Q(a,a′)=1. Then for a utility u, the general require...
b2ed35be-0bc2-4ec5-bfd7-fefbc248918b
trentmkelly/LessWrong-43k
LessWrong
Broad Basins and Data Compression Produced As Part Of The SERI ML Alignment Theory Scholars Program 2022 Research Sprint Under John Wentworth Intro The "broad basins induce data compression" argument goes like this: If we train a neural network to memorize some data, it has to "store" all the relevant information in...
f262d5b7-05d1-45bb-86d4-a2ac3d7fa5a2
trentmkelly/LessWrong-43k
LessWrong
Numeracy neglect - A personal postmortem My failed enlightenment I've been thinking about my intellectual education, and what I wish had gone differently. I am 26 years old. I’ve been reading books and going to school since I was 8. This puts my career as a learner at about 19 years. Honestly? I feel a bit disappoint...
f998728e-8688-4154-9173-ec6617195030
trentmkelly/LessWrong-43k
LessWrong
MIRI's May 2024 Newsletter Update (5-15-2024): I wrote that “it appears that not all of the leading AI labs are honoring the voluntary agreements they made at [AI Safety Summit],” citing a Politico article. However, after seeing more discussion about it (e.g. here), I am now highly uncertain about whether the labs mad...
e7902e47-09a8-4121-9397-6db9dc4b06c7
trentmkelly/LessWrong-43k
LessWrong
Why I think nuclear war triggered by Russian tactical nukes in Ukraine is unlikely This is a response to Max's post Why I think there's a one-in-six chance of an imminent global nuclear war. Max's thesis is that Putin is likely to use nuclear weapons if the war in Ukraine goes badly for Russia, that the west will ret...
da56aded-2e7b-4fa7-83c9-528c072217e3
trentmkelly/LessWrong-43k
LessWrong
Where do analogies break down? At 13, I began tinkering with every random device around my house. I pulled apart hundreds of them - remote controlled cars, table fans, aquarium purifiers, and even my PS2 controller. Figuring out how the vibrate function worked on my PS2 controller was pretty awesome! I could visualize...
99ac136e-c0ea-4782-ba70-ccaae04e8875
trentmkelly/LessWrong-43k
LessWrong
Meetup : Moscow LW meetup in "Nauchka" library Discussion article for the meetup : Moscow LW meetup in "Nauchka" library WHEN: 09 September 2016 07:50:00PM (+0300) WHERE: Москва, ул. Дубининская, 20 Welcome to the next Moscow LW meetup in "Nauchka" library! Our plan: * A talk about using Bayes formula for updati...
a7e168ff-7d4c-492c-98c5-70642518c463
trentmkelly/LessWrong-43k
LessWrong
Predictions on when the requirement for pre-flight COVID testing for returning citizen will be dropped? Did some quick searches but not getting any hits back. Anyone know if anyone is making predictions -- or just discussing -- the idea that citizens of a country will not be required to test to return home. Basically...
f2f45091-8cc7-4b83-bb30-6a108c53e3bd
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Any writeups on GPT agency? I'm searching for any articles that explore capabilities of GPT as an agent. I don't seem to find a lot (or any for that matter). Am I bad at searching?
873a0e94-e25b-49d4-b043-68fbd6988c0d
trentmkelly/LessWrong-43k
LessWrong
Map & Territory as PDF and org-mode file (& explanation) Sequences have been coming up a lot lately (have you read them?, pre-posting reqs, re-runs, and exercises). I received fantastic feedback to a recent request for advice, and the top suggestion that really solidified for me was to finish reading a set quantity of...
5c1d9098-f4a1-423a-8c13-9b4c34df5faa
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Critiquing "What failure looks like" I find myself somewhat confused as to why I should find Part I of “[What failure looks like](https://www.alignmentforum.org/posts/HBxe6wdjxK239zajf/what-failure-looks-like)” (hereafter "WFLL1", like the [pastry](https://www.google.com/search?q=waffle&oq=waffle&aqs=chrome..69i57j0l5...
a3b19815-7b8f-4dbb-95d2-65ab2356e3a3
trentmkelly/LessWrong-43k
LessWrong
Open thread, Oct. 17 - Oct. 23, 2016 If it's worth saying, but not worth its own post, then it goes here. ---------------------------------------- Notes for future OT posters: 1. Please add the 'open_thread' tag. 2. Check if there is an active Open Thread before posting a new one. (Immediately before; refresh the ...
0a0a6728-86b9-4cf4-afc5-1442b8f1bda4
trentmkelly/LessWrong-43k
LessWrong
Activation adding experiments with llama-7b Produced as part of the SERI ML Alignment Theory Scholars Program - Summer 2023 Cohort Following my initial investigation with activation adding experiments with FLAN-T5 I decided to move on to a bigger, decoder-only model (llama-7b) to see whether the results (concepts com...
2139676a-744e-41bf-a0ba-41cd24a71701
trentmkelly/LessWrong-43k
LessWrong
Public Service Announcement Collection P/S/A:  There are single sentences which can create life-changing amounts of difference. * P/S/A:  If you're not sure whether or not you've ever had an orgasm, it means you haven't had one, a condition known as primary anorgasmia which is 90% treatable by cognitive-behavioral t...
64e3e5b7-9bdc-42be-a13d-3a140d6422a0
trentmkelly/LessWrong-43k
LessWrong
To every people according to their language Here is an approximate transcript of a conversation I had with a friend, who is an Orthodox Jew. Transcript ME: Imagine the US government announced a Manhattan project to reverse aging and end death forever. It's estimated this project will cost 5% of GDP. There's no guara...
a510125b-2310-4c83-9c76-16cac4745c1b
trentmkelly/LessWrong-43k
LessWrong
Value Impact Being on Earth when this happens is a big deal, no matter your objectives – you can't hoard pebbles if you're dead! People would feel the loss from anywhere in the cosmos. However, Pebblehoarders wouldn't mind if they weren't in harm's way. Appendix: Contrived Obje...