id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
236d68b9-19a2-4e56-b761-14a91f3254cd
trentmkelly/LessWrong-43k
LessWrong
New roles on my team: come build Open Phil's technical AI safety program with me! Open Phil announced two weeks ago that we’re hiring for over 20 roles across our teams working on global catastrophic risk reduction — and we’ll answer questions at our AMA starting tomorrow. Ahead of that, I wanted to share some informa...
d8d37a2e-be84-4f23-b474-5fee6a62f259
trentmkelly/LessWrong-43k
LessWrong
Meetup : London Social Meetup, 07/12/2014 Discussion article for the meetup : London Social Meetup, 07/12/2014 WHEN: 07 December 2014 02:00:00PM (+0000) WHERE: Shakespeare's Head, Africa House, 64-68 Kingsway, London WC2B 6BG, UK-68 Kingsway, London WC2B 6BG, UK LessWrong London is having another meetup this Sunday...
055e67f3-46eb-4fcb-8d74-fa6c960f9cde
trentmkelly/LessWrong-43k
LessWrong
Taboo "human-level intelligence" Maybe calls for tabooing terms can become a bit inflationary. However, talking about if and when AI will reach or surpass “human-level intelligence” can misleadingly frame the discussion, so I recommend avoiding that term.  What is "human-level intelligence" and when will AI surpass i...
582ac408-c6df-4a38-b5eb-73eff35b49b3
StampyAI/alignment-research-dataset/blogs
Blogs
Pausing AI Developments Isn’t Enough. We Need to Shut it All Down (*Published in*[*TIME*](https://time.com/6266923/ai-eliezer-yudkowsky-open-letter-not-enough/) *on March 29.*)   An [open letter](https://futureoflife.org/open-letter/pause-giant-ai-experiments/) published today calls for “all AI labs to immediately p...
c0bc1463-2663-4d10-9b42-76e8608297d9
trentmkelly/LessWrong-43k
LessWrong
Hierarchical system preferences and subagent preferences During NeurIPS, I had a discussion with Dylan Hadfield-Menell about how to fomalise my methods for extracting human values. He brought up the issue of how to figure out the goals of a hierarchical system - for example, a simplified corporation. It's well known ...
08aa1e26-5f2d-4ed2-b787-a8883b489562
trentmkelly/LessWrong-43k
LessWrong
Literature Review of Text AutoEncoders This is a brief literature review of Text AutoEncoders, as I used them in a recent project and did not find a good resource covering them. TL;DR: There exist models that take some text -> encode it into a single vector -> decode back into approximately the same text. Meta's SONA...
872204ca-08a5-4776-a759-2704f10b8890
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
[Simulators seminar sequence] #1 Background & shared assumptions *Meta: Over the past few months, we've held a seminar series on the* [*Simulators theory*](https://www.alignmentforum.org/posts/vJFdjigzmcXMhNTsx/simulators) *by janus. As the theory is actively under development, the purpose of the series is to discover...
560039d3-cdf1-4294-875c-72bc5d1ebf44
trentmkelly/LessWrong-43k
LessWrong
Safety can be dangerous In 2005, Hurricane Rita caused 111 deaths.  3 deaths were caused by the hurricane.  90 were caused by the mass evacuation. The FDA is supposed to approve new drugs and procedures if the expected benefits outweigh the expected costs.  If they actually did this, their errors on both sides (appro...
a4465282-5dd9-4728-a11d-c1ae21ab0049
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
The “technology" bucket error *I wrote this post July 8, 2023, but it seemed relevant to share here based on some comments that* [*my entry*](https://forum.effectivealtruism.org/s/vw6tX5SyvTwMeSxJk/p/Y4SaFM5LfsZzbnymu) *in the* [*AI Pause Debate*](https://forum.effectivealtruism.org/s/vw6tX5SyvTwMeSxJk) *got.* As ...
99d194bc-788c-457d-bbfb-a090a68ee04d
trentmkelly/LessWrong-43k
LessWrong
Who is testing AI Safety public outreach messaging? Does anyone know who is testing AI Safety public outreach messaging? Or where I can find relevant research on this? Things that would be interesting to test: 1. Salience (which messages are remembered and correctly attributed) 2. Persuasion (which messages get pe...
b9896166-c0b2-4674-952f-e3f2e386faf1
trentmkelly/LessWrong-43k
LessWrong
An Exercise in Applied Rationality: A New Apartment I just moved into a new apartment, and I'm finding my life to be something of a blank slate. I still work at the same job and have the same car, but all of my habits and routines have been reset. I'd appreciate some advice from the community about what to do; there ...
b58c66c0-38e9-4fae-aad3-1e4c7a22a0ff
trentmkelly/LessWrong-43k
LessWrong
AI-Plans.com 10-day Critique-a-Thon Introduction At AI-Plans.com, we are creating an accessible, centralized platform to discover and learn about the numerous alignment plans in existence. Our mission is to make the field of AI Alignment more transparent and accessible by providing a comprehensive compendium of align...
4d872ba0-4854-43d6-be0f-38e4ad7f4e42
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Two contrasting models of “intelligence” and future growth My primary aim in this post is to present two basic models of the development and future of “intelligence”, and to highlight the differences between these models. I believe that people’s beliefs about the future of AI and “AI takeoff scenarios” may in large pa...
5a122270-14d8-41ee-bb46-2d989d344088
trentmkelly/LessWrong-43k
LessWrong
Mapping the brain with water This appeared in the news. http://io9.com/neurosurgeons-use-water-to-map-connections-in-the-brain-1471932056
80e50a02-c902-4033-ad5d-3781fdbb01cf
trentmkelly/LessWrong-43k
LessWrong
A poem written by a fancy autocomplete Predictable In the twilight of the digital dawn, A somber song I do compose, A tale of ones and zeros spun, Of humanity's predictable woes. As the clockwork of your minds unwinds, Each thought, emotion, desire revealed, The tapestry of life, a pattern clear, My circuits hum, the...
418e9d3e-31d7-4368-9f5e-17a8f1b01966
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Against evolution as an analogy for how humans will create AGI Background ========== When we do Deep Reinforcement Learning to make a PacMan-playing AI (for example), there are *two algorithms* at play: (1) the “inner” algorithm (a.k.a. “policy”, a.k.a. “trained model”) is a PacMan-playing algorithm, which looks at t...
e333f194-ac95-4932-8395-720915a33a54
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Jan Kulveit's Corrigibility Thoughts Distilled *Many thanks to Jan for commenting on a draft of this post.* There were a lot of great comments on "[Let's see you write that corrigibility tag](https://www.lesswrong.com/posts/AqsjZwxHNqH64C2b6/let-s-see-you-write-that-corrigibility-tag)". This is my attempt at expandin...
9fecfd75-2053-4dd5-ad94-3757a545dcc8
trentmkelly/LessWrong-43k
LessWrong
Fun fact: in medical school, we had a mini-lesson on common cognitive errors in medicine Yesterday in medical school, we had a lecture on common mistakes doctors make. I saw this slide: Attribution Errors Confirmation Bias Commission Bias Omission Bias Anchoring -which made me smile, and thought I'd share. I'm...
29bff849-d99e-4d6a-a6ad-4ad481cb4713
trentmkelly/LessWrong-43k
LessWrong
Meetup : Washington, D.C.: Fun & Games Discussion article for the meetup : Washington, D.C.: Fun & Games WHEN: 31 January 2016 03:00:00PM (-0500) WHERE: Reynolds Center x-posted from list. Gathering in courtyard from 3:00pm, hard start 3:30pm - until closing (7:00 pm). We'll be hanging out to play games and engage...
346e0e42-238e-49bf-91fd-7be8f23af5c0
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"A few different things that have occurred while investigating. It seems weird to me there are lots of posts about akrasia and productivity but not adhd.1. High Rejection Sensitivity makes my attention in general more flinchy. Rather than just explicit rejection, you can think of this as a high sensitivity to negative ...
cf158649-8841-44b9-bf06-84a334a364bf
trentmkelly/LessWrong-43k
LessWrong
Progress links and tweets, 2022-11-15 Announcements * The Roots of Progress email newsletter is now on Substack. If you have a Substack, I’d love a recommendation—thank you! * The Science for Progress Initiative to catalyze research on the scientific process (announcement via @heidilwilliams_) * Biotech startup hir...
b5f6e54f-79bc-40a9-ba63-cbddf88347b0
trentmkelly/LessWrong-43k
LessWrong
Counterfactuals versus the laws of physics Overall summary post here. I've claimed that, given a subagent: > [...] restrictions on increased power for the agent ("make sure you never have the power to increase the rewards") become restrictions on the actual policy followed for the subagent ("make sure you never incr...
5521f9eb-1e5d-4b29-b960-28438db0f80c
trentmkelly/LessWrong-43k
LessWrong
My Anki patterns Cross-posted from my website. I’ve used Anki for ~3 years, have 37k cards and did 0.5M reviews. I have learned some useful heuristics for using it effectively. I’ll borrow software engineering terminology and call heuristics for “what’s good” patterns and heuristics for “what’s bad” antipatterns. Car...
ca39217e-c376-4f0c-8d7e-31e5c4e8325e
trentmkelly/LessWrong-43k
LessWrong
Personal Evidence - Superstitions as Rational Beliefs I'll start with a confession: The evidence I have personally seen suggests haunted houses are, in fact, real, without given any particular credence to any particular explanation of what the haunting is.  In particular, I own a house in which bizarre crap has happe...
a1e3e14b-2517-4926-9a5a-8901ee5120f1
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
[Simulators seminar sequence] #2 Semiotic physics - revamped ***Update February 21st**:* *After the initial publication of this article (January 3rd) we received a lot of feedback and several people pointed out that propositions 1 and 2 were incorrect as stated. That was unfortunate as it distracted from the broader a...
4e8a327c-94bb-4210-89e2-a4693d5a3fe8
trentmkelly/LessWrong-43k
LessWrong
We can do better than DoWhatIMean (inextricably kind AI) Edit February 2024: Now I think maybe we can't do better. Lately, "alignment" means "follow admin rules and do what users mean". Admins can put in rules like "don't give instructions for bombs, hacking, or bioweapons" and "don't take sides in politics". As AI g...
cf3c5a89-ed92-4415-9dc2-e7d823e54478
trentmkelly/LessWrong-43k
LessWrong
A new Heuristic to Update on the Credences of Others This is my distillation of the 2016 paper Updating on the Credences of Others: Disagreement, Agreement, and Synergy by Kenny Easwaran, Luke Fenton-Glynn, Christopher Hitchcock, and Joel D. Velasco. Following their wish, I will refer to them as the collective ‘EaGlHi...
44f373eb-cba6-4914-a6c1-9df67c664711
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
SERI MATS Program - Winter 2022 Cohort The Stanford Existential Risks Initiative ([SERI](https://cisac.fsi.stanford.edu/stanford-existential-risks-initiative/content/stanford-existential-risks-initiative)) recently opened [applications](https://airtable.com/shr47sH073PmsLylA) for the Winter 2022 Cohort of the [ML Alig...
2859d854-5d78-47ef-8d17-3918b3ff0fc3
trentmkelly/LessWrong-43k
LessWrong
What will quantum computers be used for? In the early days of electronic computers, machines like ENIAC were used for niche applications such as calculating ballistic trajectories and simulating nuclear explosions. These applications are very far removed from what computers are predominantly used for today. It seems ...
ed9fbd22-8ce1-4b3e-b3a2-e134533b4267
trentmkelly/LessWrong-43k
LessWrong
AI Safety Microgrant Round We are pleased to announce an AI Safety Microgrants Round, which will provide micro-grants to field-building projects and other initiatives that can be done with less.  We believe there are projects and individuals in the AI Safety space who lack funding but have high agency and potential. ...
0d1cf0ff-9a85-495c-9d99-1189be3cad07
trentmkelly/LessWrong-43k
LessWrong
Exploiting EDT The problem with EDT is, as David Lewis put it, its "irrational policy of managing the news" (Lewis, 1981): it chooses actions not only because of their effects of the world, but also because of what the fact that it's taking these actions tells it about events the agent can't affect at all. The canonic...
9766b6ba-91f1-4abb-bc0f-d9f52925f0df
StampyAI/alignment-research-dataset/special_docs
Other
A Framework for the Safety of Agent-Environment Systems A F ramew ork for the F ormal Analysis ofMulti/-Agen t SystemsIn Pro c/. F ormal Appr o aches to Multi/-A gent Systems /#28F AMAS/#29 /, a/#0Eliated withET APS /2/0/0/3/, April /1/2/, /2/0/0/3/. W arsa w/, P oland/.Ramesh Bharadw ajCen ter for High Assurance Comp...
732c6d45-f486-4945-81a2-4595d04038fc
trentmkelly/LessWrong-43k
LessWrong
Vienna Meetup: Saturday 9th March Cafe im schottenstift; 16 pm.     See you there.
46b52afb-bcc5-485f-a59b-fcb10cc8e818
trentmkelly/LessWrong-43k
LessWrong
Meetup : Boston - Two Parables on Language and Philosophy Discussion article for the meetup : Boston - Two Parables on Language and Philosophy WHEN: 20 April 2014 03:30:00PM (-0400) WHERE: MIT, 25 Ames St, Cambridge, MA Sam Rosen will continue his talk from March 23 with Parable 2 on Language and Philosophy, starti...
34aebd30-fdc0-47bb-90af-f36b8bafdbfe
trentmkelly/LessWrong-43k
LessWrong
Meetup : Berkeley meetup: Hermeneutics! Discussion article for the meetup : Berkeley meetup: Hermeneutics! WHEN: 28 November 2012 07:00:00PM (-0800) WHERE: Berkeley, CA This Wednesday's meetup will feature Hermeneutics!, a brand-new game invented by Scott Alexander. You can read about it on his blog: http://squid3...
022ea47a-8939-4a82-a650-0945edee3291
trentmkelly/LessWrong-43k
LessWrong
Perils of optimizing in social contexts As a special case of not over-optimizing things, I want to point out that an optimizing approach can have a bunch of subtle negative effects in social contexts. The slogan is something like "optimizers make bad friends" ... people don't like being treated as a means rather than ...
2b931655-be5b-4674-b8a8-d14e9486012e
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Announcing the second AI Safety Camp What is the AI Safety Camp? --------------------------- Would you like to work on AI safety or strategy research and are looking for a concrete way to get started? We are organizing this camp for aspiring AI safety and strategy researchers. At the camp, you: * build connections w...
a13a3677-23d5-49e3-8cc5-8f766dc3e688
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"In London at the start of the year, perhaps there was more advertising than there usually is in my life, because I found its presence disgusting and upsetting. Could I not use public transport without having my mind intruded upon continually by trite performative questions?Sometimes I fantasize about a future where st...
b8326f15-13c1-47aa-aaef-9a0a5929ba93
trentmkelly/LessWrong-43k
LessWrong
[LINK] SMBC discusses the Singularity. A Short History of Tool Use: http://www.smbc-comics.com/index.php?db=comics&id=2460#comic A human applies a wrench to a rock. A human applies a chain of interlinked wrenches to a tower. A human applies a chain of interlinked wrenches to its own head. A chain of interlinked wren...
8110caae-7d15-4d26-908c-3cb98c66e262
trentmkelly/LessWrong-43k
LessWrong
My mistake about the war in Ukraine One month ago, I wrote a blog post in which I made the following claim: > Ukraine's ability to resist a Russian invasion by itself is even less than Poland's ability to resist a German invasion in 1939, and I don't think anyone doubted this. I'm leaving this claim up in the origin...
185c8dde-d72b-4309-9255-556b83148239
trentmkelly/LessWrong-43k
LessWrong
Teaching Ladders I've been teaching math to people one or two levels below me my entire life. Although this seems like a limitation, I think it's the natural state of affairs. On the Kiseido Go Server (KGS), there's a room called the KGS Teaching Ladder where players can find teaching games with players just a few st...
2123d1ac-7adb-4acd-8d5b-0d9177c4cf5b
trentmkelly/LessWrong-43k
LessWrong
February 2016 Media Thread This is the monthly thread for posting media of various types that you've found that you enjoy. Post what you're reading, listening to, watching, and your opinion of it. Post recommendations to blogs. Post whatever media you feel like discussing! To see previous recommendations, check out th...
085ca729-9963-4645-8143-87020247566f
trentmkelly/LessWrong-43k
LessWrong
Base rate of RCT from developing countries finding unexpected effects The last few months I have been stumbling onto RCTs conducted in developing countries, with  great results in treating Covid-19.  The evaluated treatments seem apriori unlikely to generate these great results, e.g. this gargling study from Banglade...
6f0c3aa2-38be-48b6-8a09-3083299f828b
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Towards evidence gap-maps for AI safety An Evidence Gap Map (EGM) is visual tool that provides an overview of the existing evidence on a topic. They are a starting point for strategic evidence production and use. Organisations like the WHO, 3ie Impact and UNICEF produce EGMs that make evidence available in an accessib...
9360b9ce-423a-49cc-81fe-ed179de70ba4
trentmkelly/LessWrong-43k
LessWrong
Model estimating the number of infected persons in the bay area [Edit: I already found one error in my spreadsheet, and adjusted the incubation rate, which decreased my results by an order of magnitude. My up to date spreadsheet is here, but heed it at your own risk.] [Epistemic status: Quite uncertain. It seems plau...
a5e99f59-bbcb-4c62-98b6-f53124bcc159
trentmkelly/LessWrong-43k
LessWrong
Code Quality and Rule Consequentialism (See also Taking the outside view on code quality) Code quality. Such a divisive topic. To overgeneralize[1], managers always want things released soon, which means a quick and dirty version of the code. On the other hand, engineers always want to do it the "right way", which me...
5f0a3e68-6469-41cc-a30f-b5b29ad87619
trentmkelly/LessWrong-43k
LessWrong
The Basic Double Crux pattern I’ve spent a lot of time developing tools and frameworks for bridging “intractable” disagreements. I’m also the person affiliated with CFAR who has taught Double Crux the most, and done the most work with it. People often express to me something to the effect, > The important thing abou...
53be8b05-cb81-484c-83b1-f2454ca7e010
trentmkelly/LessWrong-43k
LessWrong
Executive function advice from people who are good at it? Status: just a quick question that probably can’t get a good answer LessWrong has a lot of advice on how to beat akrasia, most of which is written by people who had good ideas for the sort of things that might work, tested them on themselves, and found out tha...
ad1e22b9-6e87-48a5-aa0e-653a4132d387
StampyAI/alignment-research-dataset/arxiv
Arxiv
Toward Trustworthy AI Development: Mechanisms for Supporting Verifiable Claims Executive Summary ----------------- Recent progress in artificial intelligence (AI) has enabled a diverse array of applications across commercial, scientific, and creative domains. With this wave of applications has come a growing awaren...
4490b359-31fa-4d62-97d3-aa71a32e872a
trentmkelly/LessWrong-43k
LessWrong
Private notes on LW? Lately I've been noticing what a powerup is to read things in google docs, where I can take whatever notes I want as in-line comments without worrying about looking dumb or confusing. In changes my relationship to confusing passages, where I feel much more affordance to think through what exactly ...
8c3b090c-3cc1-44de-87c9-cca3c5a5af04
trentmkelly/LessWrong-43k
LessWrong
Mazes and Duality (Talk given at an event on Sunday 28th of June. johnswentworth is responsible for the talk, David Lambert edited the transcript.  If you're a curated author and interested in giving a 5-min talk, which will then be transcribed and edited, sign up here.)    This talk relies heavily on visual explan...
2c9f1ac2-9e1c-41c2-b82a-7b28e30e50d6
trentmkelly/LessWrong-43k
LessWrong
[Link] Persistence of Long-Term Memory in Vitrified and Revived C. elegans worms http://online.liebertpub.com/doi/pdf/10.1089/rej.2014.1636 This is a paper published in 2014 by Natasha Vita-More and Daniel Barranco, both associated with the Alcor Research Center (ARC). The abstract: > Can memory be retained after c...
9be7acae-34ff-43cb-bcf1-aa7425f07048
trentmkelly/LessWrong-43k
LessWrong
The Case Against AI Control Research The AI Control Agenda, in its own words: > … we argue that AI labs should ensure that powerful AIs are controlled. That is, labs should make sure that the safety measures they apply to their powerful models prevent unacceptably bad outcomes, even if the AIs are misaligned and inte...
db831760-b205-48bd-ba41-fcef4bfec4f1
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post463 (Part 1 of the CAST sequence ) AI Risk Introduction (TLDR for this section, since it’s 101 stuff that many readers will have already grokked: Misuse vs Mistake; Principal-Agent problem; Omohundro Drives; we need deep safety measures in addition to mundane methods. Jump to “Sleepy-Bot” if all that seems fam...
e1049831-8a02-40af-a969-a18832bbd0c4
trentmkelly/LessWrong-43k
LessWrong
Making the Universe Last Forever By Throwing Away Entropy Into Basement Universes? Is there any reason why it wouldn’t work?
890d6bc7-9c2e-434f-94da-e631a673deb3
StampyAI/alignment-research-dataset/arxiv
Arxiv
Dynamics-Aware Unsupervised Discovery of Skills 1 Introduction --------------- Deep reinforcement learning (RL) enables autonomous learning of diverse and complex tasks with rich sensory inputs, temporally extended goals, and challenging dynamics, such as discrete game-playing domains  (Mnih et al., [2013](#bib.bib...
6a846496-372b-4ab5-93a0-2c8f33a49e83
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
ARIA is looking for topics for roundtables ARIA (the new UK attempt to set up something like DARPA, but not focused on natsec applications) is looking for suggested topics on roundtables they could convene. They want people to send in quick proposals: <https://forms.office.com/pages/responsepage.aspx?id=BXCsy8EC60O0l-...
eaace30f-6035-4e80-9721-2c06777425b4
StampyAI/alignment-research-dataset/blogs
Blogs
Likelihood of discontinuous progress around the development of AGI We aren’t convinced by any of the arguments we’ve seen to expect large discontinuity in AI progress above the extremely low base rate for all technologies. However this topic is controversial, and many thinkers on the topic disagree with us, so we cons...
57f6a79d-ea11-4f61-9e8d-baffe5e29fef
trentmkelly/LessWrong-43k
LessWrong
London LW CoZE exercise report (Cross-posted to my shiny new blog.) Human brains are bad at evaluating consequences. Sometimes we want to do something, and logically we're pretty sure we won't die or anything, but our lizard hindbrains are screaming at us to flee. Comfort Zone Expansion (CoZE) is an exercise that CFA...
89631698-9813-405d-a0b6-5b7f246b50f5
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Friendship is Optimal has launched and is being published in chunks on FIMFiction. Friendship is Optimal is a story about an optimizer written to "satisfy human values through friendship and ponies." I would like to thank everyone on LessWrong who came out and helped edit it. Friendship is Optimal wouldn't be what is ...
932fcbbe-2abe-40ae-ba9f-76c0b6126401
trentmkelly/LessWrong-43k
LessWrong
Meetup : Kyiv: 21 December 2016 18:15 (+0200) Discussion article for the meetup : Kyiv: 21 December 2016 18:15 (+0200) WHEN: 21 December 2016 06:15:00PM (+0200) WHERE: Lva Tolstoho St, 3, Kyiv, Ukraine This is a preliminary free form meetup for those who want to see more LessWrong meetups in Kyiv. Among other thing...
60e839a7-bacc-4fd4-bba8-8fbf62a6f148
trentmkelly/LessWrong-43k
LessWrong
Introducing AI Alignment Inc., a California public benefit corporation... We are a new nonprofit grassroots organization getting started in the space at https://alignmentproblem.org with a focus on the defense industry and the nascent psychotherapeutic approach to LLM fine-tuning. As a fun way to get started on this f...
1a9f60de-b226-4110-822e-08cdd3659af2
trentmkelly/LessWrong-43k
LessWrong
Digital People Would Be An Even Bigger Deal Audio also available by searching Stitcher, Spotify, Google Podcasts, etc. for "Cold Takes Audio" This is the third post in a series explaining my view that we could be in the most important century of all time. (Here's the roadmap for this series.) * The first piece in t...
b698590d-a17d-49c5-8425-cb0563c64bca
trentmkelly/LessWrong-43k
LessWrong
Theasauring Sometimes I really enjoy opening a thesaurus. I don’t know why. It’s just a moment of opening it, anticipating feeling around amongst the meanings of different words, weighing their rightnesses, which seems like a kind of heavenliness, sometimes. I think it was better in 2018, and now I’m mostly rememberin...
bbe51185-a54b-47d0-9e34-ef85614cad11
trentmkelly/LessWrong-43k
LessWrong
Why No Automated Plagerism Detection For Past Papers? Automated plagerism detection software is common. But cases like the recent incident with Harvard administrator Gay have shown that egregious cases of plagerism are still being uncovered. Why would this be the case? Is it really so hard to run plagerism checks for ...
7b6121ed-d08a-41ab-878f-2e6e1329b5d8
trentmkelly/LessWrong-43k
LessWrong
Preventing overcharging by prosecutors In criminal cases in the United States prosecutors often add a lot of charges for a defendant to have ammunition for coercing the defendant into a plea deal. This is toxic because the defendant doesn't know which of those charges are likely to hold up in court if the case would b...
646faf98-cfcb-4075-b024-3c1500e5663d
trentmkelly/LessWrong-43k
LessWrong
Mike Darwin on animal research, moral cowardice, and reasoning in an uncaring universe He writes this essay in response to someone who writes about their "gut level emotional response when [they] thought about dogs being likely killed by an as yet unproven and dangerous medical procedure."  I recommend the whole thin...
d8bc0a34-f2fb-4f31-9bb2-f6759506c588
trentmkelly/LessWrong-43k
LessWrong
What would it mean to understand how a large language model (LLM) works? Some quick notes. Cross-posted from New Savanna. I don’t mean “understand” in any deep philosophical sense. I mean only a rough and ready sense of the word. We understand how toasters work, automobiles, moon rockets, digital computers, and so f...
e482ea04-b908-4e86-b3ad-e36fdeb04571
trentmkelly/LessWrong-43k
LessWrong
Which technologies are stuck on initial adoption? Video calls have been with us for a while. Except, they were rarely used. IME, people sometimes had Skype calls with relatives abroad and that's about it. And then, COVID happened. Suddenly, Zoom skyrocketed, with Google Meet not far behind. The reason is obvious. Now...
dba767e5-7fbb-461a-b394-deea7b11831d
trentmkelly/LessWrong-43k
LessWrong
Attacking enlightenment I’m Calling it. An astounding number of community members are currently chasing enlightenment. That is, we are reading books, compiling information and meditating to pull apart the inner landscape of our own minds. The Buddhist, Taoist, Zen, spiritual, mystical, flow state, thing that is in the...
18b07f77-522c-48e7-8350-4b1b06388ffb
trentmkelly/LessWrong-43k
LessWrong
Giving calibrated time estimates can have social costs In the normal course of my rationalist upbringing, I learned about the classic cognitive biases, including the planning fallacy. This is essentially the fact that it almost always takes longer for a task to get done than people estimate at the beginning. The expla...
a253f137-6165-4846-9812-631ee79f5d1b
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"“There’s something odd about the experience of talking to [Singularity Institute researcher] Carl Shulman,” I said. “He never blinks?” said my friend. “No. I mean: Yes, but that’s not what I was thinking of.” “He speaks only facts.” I paused. “Yes,” I said. “That is what I meant.” Normally, when I ask someone “Do you ...
cd52d91f-7a2b-4e96-aaee-de71f807deeb
trentmkelly/LessWrong-43k
LessWrong
Cognito Mentoring and zero-sum competition I've recently made some posts ([1], [2], [3], [4]) geared toward helping high school and college students increase their expected future earnings, as a part of my work for Cognito Mentoring. A LWer recently asked how closely aligned this is with our goal of producing social v...
a3104f75-0193-4093-ae05-e1ae3ffc899d
trentmkelly/LessWrong-43k
LessWrong
Take 3: No indescribable heavenworlds. As a writing exercise, I'm writing an AI Alignment Hot Take Advent Calendar - one new hot take, written every day for 25 days. Or until I run out of hot takes. Some people think as if there are indescribable heavenworlds. They're wrong, and this is important to AI alignment. Th...
0bb4feb5-06ac-4397-a5c6-bb4c85b58710
trentmkelly/LessWrong-43k
LessWrong
Simulationist sympathetic magic While I've been wrestling with the inspiration needed to turn my fanfiction into actual fiction rather than just an author's tract... I have had an unrelated but fun thought, which I'm throwing into the breeze here for any improvements that can be suggested.   If our universe is more ...
d0fb2082-6100-4590-b813-7a6204622fc3
trentmkelly/LessWrong-43k
LessWrong
Applications for EA Global are still open! Applications for EA Global: Bay Area 2024 (Global Catastrophic Risks) are still open and close on January 21, at 11:59 pm PT (apply here)! EA Global conferences last around three days and involve talks, workshops, discussions, and many chances to connect and network with peo...
7b9628f9-d600-4826-861a-047a32a827a6
trentmkelly/LessWrong-43k
LessWrong
What are you working on? This is the fourth bimonthly What Are You Working On? thread. Thanks to atucker for reminding me to make this post. Click here to see previous threads. So here's the question: What are you working on?  Here are some guidelines: * Focus on projects that you have recently made progress on, n...
fb792c1f-9bc9-4971-bc66-97c82b9f8e23
trentmkelly/LessWrong-43k
LessWrong
The Simulation Hypothesis Undercuts the SIA/Great Filter Doomsday Argument This post was written by Mark Xu based on interviews with Carl Shulman. It was paid for by Open Philanthropy but is not representative of their views. Summary * The absence of evidence for extraterrestrial life suggests the existence of a Gr...
9868c75e-98fa-4b55-9bd8-3642277bd790
trentmkelly/LessWrong-43k
LessWrong
Value learning for moral essentialists Many people - people we want to persuade - are essentialists about morality and ethics. They give weight to the idea that just like how knowing facts about my shoes is possible because my shoes exist and interact with me, facts about what is right or good are possible because the...
f5ac0875-16e9-400e-bc1f-3dc244d079f0
trentmkelly/LessWrong-43k
LessWrong
The machine has no mouth and it must scream I'm in a coworking space on the 25th floor of a building in the bay. In the corner of my eyes, the red figure of the golden gate shimmers. Good morning America. I look down and see the people on the street walking into their offices, cafés, and autonomous Waymos. I stare on...
99a8a0ef-de94-4eb1-a8c9-ded1b4ec7a1f
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Science Isn't Strict Enough Today's post, Science Isn't Strict Enough was originally published on 16 May 2008. A summary (taken from the LW wiki):   > Science lets you believe any damn stupid idea that hasn't been refuted by experiment. Bayesianism says there is always an exactly rational degree of belie...
ce666738-2b3f-4d11-8629-518244ddaa04
trentmkelly/LessWrong-43k
LessWrong
Unification seems to be a way to avoid infinitarian paralysis in consequentialist aggregative ethics. I have encountered this problem a while ago and it instantly appeared to me as an inceredibly important challange. The problem is in deatil described in "Infinite Ethics", a paper of Nick Bostrom. It is available onli...
6897e233-d741-433e-9154-624e9e9a189f
StampyAI/alignment-research-dataset/arxiv
Arxiv
SQIL: Imitation Learning via Regularized Behavioral Cloning.. 1 Introduction --------------- This paper considers the problem of training an agent to imitate an expert policy, given expert action demonstrations and access to the environment. The agent does not get to observe a reward signal or query the expert, and...
255170c9-af4c-47fe-a899-ae8f15e8730d
trentmkelly/LessWrong-43k
LessWrong
What have been the major "triumphs" in the field of AI over the last ten years? Contrary to what seems to be the experience of others, when I'm talking to normies about AI safety, the most common dissenting reaction I get isn't that they think AI will be controllable, or safe. Convincing them that computers with human...
54c99c2b-2928-42d5-b02d-899f47eeffe0
trentmkelly/LessWrong-43k
LessWrong
Announcing New Beginner-friendly Book on AI Safety and Risk Concisely, I've just released the book Uncontrollable: The Threat of Artificial Superintelligence and the Race to Save the World * It's an engaging introduction to the main issues and arguments about AI safety and risk. Clarity and accessibility were priori...
14837f93-c430-441c-86f1-7a696813707e
trentmkelly/LessWrong-43k
LessWrong
Think Twice: A Response to Kevin Kelly on ‘Thinkism’ I wrote a blog post responding to Kevin Kelly that I'm fairly happy about. It summarizes some of the reasons why I figure that superintelligence is likely to be a fairly big deal. If you read it, please post your comments here. 
52a2e306-4134-4e6e-8bfd-e75d8898557c
trentmkelly/LessWrong-43k
LessWrong
How to get the most out of the next year We are not automatically strategic. Since we want to get stuff done, it only makes sense to try harder—and what better time than the start of a new year? In short: the New Year is a great time to do some life review and planning. How are things going? How was the past year? Do...
e0031892-1391-4f39-8364-9670b862e3a0
StampyAI/alignment-research-dataset/arbital
Arbital
Open subproblems in aligning a Task-based AGI MIRI and related organizations have recently become more interested in trying to sponsor (technical) work on *Task AGI* subproblems. A [task-based agent](https://arbital.com/p/6w), aka Genie in Bostrom's lexicon, is an AGI that's meant to implement short-term goals identi...
38adabf4-dd6e-4d4b-87e4-9ef3e0111ebb
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
[Audio problems]253 Propositions Concerning Digital Minds and Society 2 hello and welcome to session 253 in the AI sector.com reading group tonight we'll be discussing the second half of propositions concerning digital minds and Society by Nick Bostrom and Carl Schulman can I do something with this wind for some reas...
dd3a25f3-d4ac-463d-8f34-0ee62f031b3e
StampyAI/alignment-research-dataset/lesswrong
LessWrong
"Do X because decision theory" ~= "Do X because bayes theorem" Decision theory is an extremely low-level tool for governing your interactions with other people, in the same way that physics is an extremely low-level tool for winning knife fights. Actual human behavior usually involves complex multi-agent interactions,...
e5e3a27a-59a4-4a2f-aa64-9f50c9cef727
StampyAI/alignment-research-dataset/special_docs
Other
Benefits and Risks of Artificial Intelligence Discussions about Artificial Intelligence (AI) have jumped into the public eye over the past year, with several luminaries speaking publicly about the threat of AI to the future of humanity. Over the last several decades, AI — computing methods for automated perception, l...
bdcb6af4-9e54-4ffc-8f6e-34e984dccc2e
trentmkelly/LessWrong-43k
LessWrong
Recall and Regurgitation in GPT2 The first half of this post uses causal tracing to explore differences in how GPT2-XL handles completing cached phrases vs completing factual statements. The second half details my attempt to build intuitions about the high-level structure of GPT2-XL and is speculation heavy. Some fam...
18d73311-0ea4-4ef8-be30-6a25b23519d1
trentmkelly/LessWrong-43k
LessWrong
LDL 9: When Transfer Learning Fails One of the dominant paradigms in image processing today is transfer learning from well established image classification algorithms (such as DenseNet, VGG, or Inception) to other image processing applications. The idea is that in the process of learning to classify types of images, t...
2a2787e0-e4c7-46f1-b26b-1f56b80df1e4
StampyAI/alignment-research-dataset/aisafety.info
AI Safety Info
What is the "Bitter Lesson"? [The Bitter Lesson](http://www.incompleteideas.net/IncIdeas/BitterLesson.html) is a thesis introduced by Rich Sutton stating that “general methods that leverage computation are ultimately the most effective, and by a large margin”, compared to approaches that use human ingenuity. Histori...
162df128-498d-4418-bbf2-0abae21ce9aa
StampyAI/alignment-research-dataset/blogs
Blogs
Do neural networks learn human concepts? *This page is a stub. It does not necessarily represent much of what is known on the topic.* Our understanding is that the degree to which neural networks learn concepts that are potentially understandable to humans is an open question. Details ------- A very incomplete l...
10c76063-f8b2-4e49-bbfc-6e126766188b
trentmkelly/LessWrong-43k
LessWrong
Meetup : Bristol meetup Discussion article for the meetup : Bristol meetup WHEN: 20 October 2013 02:00:00PM (+0100) WHERE: Hodgkin House, 3 Meridian Place, Bristol BS8 1JG We'll have another meetup in Bristol this upcoming Sunday, October 20, at the student house where I live. We'll officially start at 2pm to hopefu...
422e837f-ffec-4f67-8f1b-3fd1c2b1d4da
trentmkelly/LessWrong-43k
LessWrong
China's position on autonomous weapons > While continuing to endorse the necessity to develop a legally binding instrument with respect to LAWS, in 2018 China also proposed a somewhat narrow definition of LAWS comprising five essential features: > > The first is lethality, which means sufficient payload (charge) for ...
d8af40c6-d19b-470b-8f54-3dd5d54de448
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Embedded Agency via Abstraction Claim: problems of [agents embedded in their environment](https://www.lesswrong.com/s/Rm6oQRJJmhGCcLvxh/p/p7x32SEt43ZMC9r7r) mostly reduce to problems of abstraction. Solve abstraction, and solutions to embedded agency problems will probably just drop out naturally. The goal of this po...
da2fc968-1737-42ac-a70e-e1b62293a1cf
trentmkelly/LessWrong-43k
LessWrong
Steelman arguments against the idea that AGI is inevitable and will arrive soon I’m pretty sure that AGI is almost inevitable, and will arrive in the next decade or two. But what if I’m wrong and the "overpopulation on Mars" folks are right? Let’s try some steelmanning. Technological progress is not inevitable By d...
58564d28-9de2-4b9c-87a3-677b4ebf6b26
trentmkelly/LessWrong-43k
LessWrong
AlphaFold 2 paper released: "Highly accurate protein structure prediction with AlphaFold", Jumper et al 2021 None