id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
241f2e50-8e8d-49ac-a351-483259b7a5fd
trentmkelly/LessWrong-43k
LessWrong
Great Books of Failure Followup to:  Unteachable Excellence As previously observed, extraordinary successes tend to be considered extraordinary precisely because it is hard to teach (relative to the then-current level of understanding and systematization).  On the other hand, famous failures are much more likely to c...
223461fb-c395-46d5-922e-374ac098dc51
trentmkelly/LessWrong-43k
LessWrong
Alignment Newsletter #15: 07/16/18 Highlights Feature-wise transformations (Vincent Dumoulin et al): This Distill article is about transformations on features using FiLM (feature-wise linear modulation). A FiLM layer is used to "condition" a neural network on auxiliary information, which just means providing the input...
8f4b74e6-0bbf-41b5-a46c-1bc3ec6236cd
StampyAI/alignment-research-dataset/special_docs
Other
International Cooperation vs. AI Arms Race International Cooperation vs. AI Arms Race ========================================== 8 April 2015 by [Brian Tomasik](https://longtermrisk.org/author/brian-tomasik/ "Posts by Brian Tomasik") First written: 5 Dec. 2013; last update: 29 Feb. 2016 Summary ------- There's...
526c3bde-55c6-4c71-83f1-9849553d1b01
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Tan Zhi Xuan: AI alignment, philosophical pluralism, and the relevance of non-Western philosophy --- *How can we build (super) intelligent machines that are robustly aligned with human values? AI alignment researchers strive to meet this challenge, but currently draw upon a relatively narrow set of philosophical pers...
67594593-23e6-4352-ae9c-315844cb155c
trentmkelly/LessWrong-43k
LessWrong
Two Notions of Best Response In game theory, there are two different notions of "best response" at play. Causal best-response corresponds to standard game-theoretic reasoning, because it assumes that the joint probability distribution over everyone else's moves remains unchanged if one player changes their move. The s...
4792064f-85fb-4d37-a4cb-c351208f5020
trentmkelly/LessWrong-43k
LessWrong
Can efficiency-adjustable reporting thresholds close a loophole in Biden’s executive order on AI? Epistemic Status: Exploratory President Biden’s Executive Order on the Safe, Secure, and Trustworthy Development and Use of Artificial Intelligence specifies compute thresholds for training runs and computing clusters th...
5d1685ad-f1cb-4986-85ca-2995febaec7e
trentmkelly/LessWrong-43k
LessWrong
Harsanyi's Social Aggregation Theorem and what it means for CEV A Friendly AI would have to be able to aggregate each person's preferences into one utility function. The most straightforward and obvious way to do this is to agree on some way to normalize each individual's utility function, and then add them up. But ma...
516d70f2-3245-4115-80a8-8efddc9131af
trentmkelly/LessWrong-43k
LessWrong
Three Worlds Decide (5/8) (Part 5 of 8 in "Three Worlds Collide") Akon strode into the main Conference Room; and though he walked like a physically exhausted man, at least his face was determined.  Behind him, the shadowy Confessor followed. The Command Conference looked up at him, and exchanged glances. "You look ...
9c938c3f-7bd1-4420-889a-82d9d2c19e66
StampyAI/alignment-research-dataset/aisafety.info
AI Safety Info
What is outer alignment? [Outer alignment](/?state=8428&question=What%20is%20the%20difference%20between%20inner%20and%20outer%20alignment%3F), also known as the reward misspecification problem, is the problem of training the AI with the right optimization objective, i.e., “Did we tell the AI *what to do* correctly?” T...
c0f5e52b-d6d1-49dc-996c-d5bc06e4fdef
trentmkelly/LessWrong-43k
LessWrong
Social Dilemmas — public goods, free riders, and exploitation The Stag Hunt showed that different parties can coordinate to achieve the best payoff, and we find that the situation becomes more complex with more players. In the Public Goods Game we are dealing with a large group of players who also have individual inte...
edf3b63d-d061-48cd-bd8c-16f683675890
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Elon Musk donates $10M to the Future of Life Institute to keep AI beneficial We are delighted to report that technology inventor Elon Musk, creator of Tesla and SpaceX, has decided to donate $10M to the Future of Life Institute to run a global research program aimed at keeping AI beneficial to humanity.  There is n...
919bec40-b308-4eeb-a5ca-7985407257e2
trentmkelly/LessWrong-43k
LessWrong
2021 New Year Optimization Puzzles I was asked to host a puzzle corner for the New Year Eve's Ultra Bash, which would be a Schelling time/place for people to talk about and solve puzzles together. For the event, I decided to make three 2021 themed optimization puzzles, and I decided to also share them here. For each o...
d022a42d-7d33-4503-9cea-281971ccc346
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Catastrophic Risks from AI #1: Introduction *This is the first post in a sequence of posts giving an* [*overview of catastrophic AI risks*](https://arxiv.org/abs/2306.12001)*.* Abstract ======== Rapid advancements in artificial intelligence (AI) have sparked growing concerns among experts, policymakers, and world le...
679a2e13-9677-44ea-883c-625be06209ff
StampyAI/alignment-research-dataset/lesswrong
LessWrong
What Evidence Is AlphaGo Zero Re AGI Complexity? Eliezer Yudkowsky write a post on Facebook on on Oct 17, where I replied at the time. Yesterday he reposted that here ([link](https://www.lesserwrong.com/posts/shnSyzv4Jq3bhMNw5/alphago-zero-and-the-foom-debate)), minus my responses. So I’ve composed the following resp...
8ba35acf-8887-4550-84d8-fdda906f60dd
trentmkelly/LessWrong-43k
LessWrong
ACX Meetups Everywhere 2023: Times & Places Thanks to everyone who responded to my request for ACX meetup organizers. Volunteers have arranged meetups in 169 cities around the world, from Baghdad to Bangalore to Buenos Aires. You can find the list below, in the following order: Africa & Middle East Asia-Pacific Eu...
41cc0d35-73e5-4f15-9b2d-5183cb97e93f
trentmkelly/LessWrong-43k
LessWrong
if you're reading this it's too late (a new theory on what is causing the Great Stagnation)   A sword is against the internet, against those who live online, and against its officials and wise men. A sword is against its false prophets, and they will become fools. A drought is upon its waters, and they will be dried u...
a9fd8daa-483b-4640-a936-d919288a7c72
trentmkelly/LessWrong-43k
LessWrong
The Dark Arts - Preamble I’d like to tell you all a story. Once upon a time I was working for a charity – a major charity – going door-to-door to raise money while pretending it wasn’t sales. This story happened on my last day working there.  I didn’t know that at the time; I wouldn’t find out until the following mo...
e68a7092-90a9-4996-a73b-4bdf13954499
trentmkelly/LessWrong-43k
LessWrong
SIAI - An Examination 12/13/2011 - A 2011 update with data from the 2010 fiscal year is in progress. Should be done by the end of the week or sooner.   Disclaimer * I am not affiliated with the Singularity Institute for Artificial Intelligence. * I have not donated to the SIAI prior to writing this. * I made thi...
f835a3ae-84d7-4882-a86f-6cd90173dbbe
trentmkelly/LessWrong-43k
LessWrong
Welcome to Effective Altruism Ottawa [Edit With Your Details] (The following are our suggestions for what kind of information is best to include in the welcome post of your group, feel free to replace them with whatever you think is best) What kind of events does your group usually run? What does it usually do? How ...
428d45dd-b15b-4cb1-a731-e26bc31832f3
trentmkelly/LessWrong-43k
LessWrong
Dan Hendrycks and EA According to public reports, Dan Hendrycks has been influenced by EA since he was a freshman (https://www.bostonglobe.com/2023/07/06/opinion/ai-safety-human-extinction-dan-hendrycks-cais/).  He did the 80,000 hours program.  He worries about AI bringing about the end of humanity, if not the plan...
4193dc37-3f94-477e-ae5f-7afb16f543c6
trentmkelly/LessWrong-43k
LessWrong
Repository repository A few weeks ago, Adele_L suggested that the repositories were underutilized and looked for suggestions on how to improve that. In that spirit, I added the following links to the Special Threads wiki page. Solved Problems Repository - A collection of "solved problems in instrumental rationality."...
db42dc93-840c-4522-acca-6b43319162b5
trentmkelly/LessWrong-43k
LessWrong
Is progress in ML-assisted theorem-proving beneficial? I want to believe that the outputs of the recently founded Hoskinson Center for Formal Mathematics (led by Jeremy Avigad, researcher of mathematical logic, automated reasoning and philosophy of mathematics), will live up to Mr Hoskinson's aura of technical respons...
d27c682c-faad-4d45-9a45-29ce703b87da
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
A shot at the diamond-alignment problem I think that relatively simple alignment techniques can go a long way. In particular, I want to tell a plausible-to-me story about how simple techniques can align a proto-AGI so that it makes lots of diamonds. But why is it interesting to get an AI which makes lots of diamonds?...
bcf95aed-1d33-428a-95eb-78867f3c324e
trentmkelly/LessWrong-43k
LessWrong
Individual Deniability, Statistical Honesty If you have a lot of people to question about something, and they have a motivation to lie, consider this clever use of a six-sided die. > If the farmer tossed the die and got a one, they had to respond "yes" to the surveyor's question. If they got a six, they had to say "n...
f5ebc150-77a7-4256-a6b7-7843a445f49e
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
144. Value Learning with Rohin Shah hello and welcome to session 144 in the AIC t-dot-com reading group tonight we have roving shot joining us from the Center for human compatible AI and he will answer some of the questions that we have that have come up during the discussion of the value learning sequence hello Rowan...
3b79656d-a25c-4cdf-aa96-3143a0beda6e
trentmkelly/LessWrong-43k
LessWrong
Debating myself on whether “extra lives lived” are as good as “deaths prevented” Click lower right to download or find on Apple Podcasts, Spotify, Stitcher, etc. If we could do something to lower the probability of the human race going extinct,1 that would be really good. But how good? Is preventing extinction more...
9bab8adb-2af3-4a9a-ae0a-f55646926d2f
trentmkelly/LessWrong-43k
LessWrong
Fiction of interest The fiction piece in this week's New Yorker deals with some of the same themes as Eliezer's "Three Worlds Collide"; viz., the clash of value systems (and the difficulty of seeing those with a different value system as rational), and the idea of humanity developing in ways that seem bizarre/grotesqu...
37bb16c8-ac9f-4a4c-9de3-a2b5da97d158
trentmkelly/LessWrong-43k
LessWrong
the micro-fulfillment cambrian explosion Warehouse automation has been very successful. Here's a typical modern system. As you can see, a tall narrow robot rides on a single rail at the top and bottom. The linked example is up to 42m tall. Items are stored on top of pallets, and the robot has a telescoping fork, which...
5fd22622-42dc-476a-8896-6907eb8afd2c
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Riffing on the agent type *Much is owed to Diffractor for Giry-pilling me at Alignable Structures, I had been struggling with type-driven expected value previously.* *Epistemic status: took a couple days off from my [master plan](https://www.lesswrong.com/posts/HbPXiB5Aiv3uHPXEh/master-plan-spec-needs-audit-logic-an...
8700a39e-c618-492e-bc52-df27b58f9b64
trentmkelly/LessWrong-43k
LessWrong
Progress links and short notes, 2025-02-17 Much of this content originated on social media. To follow news and announcements in a more timely fashion, follow me on Twitter, Notes, Farcaster, Bluesky, or Threads. Contents * My writing (ICYMI) * Job opportunities * Funding opportunities * Tech announcements * Wri...
cd34f2b2-8ed0-4378-834d-04a34eaa13c6
trentmkelly/LessWrong-43k
LessWrong
Three Types of Intelligence Explosion Abstract Once AI systems can design and build even more capable AI systems, we could see an intelligence explosion, where AI capabilities rapidly increase to well past human performance. The classic intelligence explosion scenario involves a feedback loop where AI improves AI sof...
8d878663-3823-4b72-b707-050415cee493
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Inseparably Right; or, Joy in the Merely Good Today's post, Inseparably Right; or, Joy in the Merely Good was originally published on 09 August 2008 . A summary (taken from the LW wiki):   > Don't go looking for some pure essence of goodness, distinct from, you know, actual good. Discuss the post here ...
82fbba46-435e-41af-a31c-dde5b061dc2c
trentmkelly/LessWrong-43k
LessWrong
Analysis of World Records in Speedrunning [LINKPOST] [this is a linkpost to Analysis of World Records in Speedrunning] ETA: a more recent analysis can be found here. TL;DR: I have scraped a database of World Record improvements for fastest videogame completition for several videogames, noted down some observations a...
4db826bd-9f62-4999-83da-ea14cdced997
trentmkelly/LessWrong-43k
LessWrong
The 9/11 Meta-Truther Conspiracy Theory Date:  September 11th, 2001. Personnel:  Unknown [designate A], Unknown [designate B], Unknown [designate C]. A:  It's done.  The plane targeted at Congress was crashed by those on-board, but the Pentagon and Trade Center attacks occurred just as scheduled. B:  Congress seems ...
86713c94-2e65-47f9-8edd-e753c59c36d7
trentmkelly/LessWrong-43k
LessWrong
LINK: Can intelligence explode? I thought many of you would be interested to know that the following paper just appeared in Journal of Consciousness Studies: "Can Intelligence Explode?", by Marcus Hutter. (LINK HERE) Abstract: The technological singularity refers to a hypothetical scenario in which technological adv...
2ccbe681-d1ba-46f3-940e-d695bf71bf3d
trentmkelly/LessWrong-43k
LessWrong
House Covid Infection Retrospective On Saturday 2023-01-14 our 6yo tested positive for covid. Three weeks later we're all fine, but five of us ended up getting it despite relatively strong efforts to isolate. In retrospect I think this wasn't worth it, and even though I didn't get sick I would rather have gotten covid...
0ee24ed9-16fc-43c4-8e50-9a58a871b36f
trentmkelly/LessWrong-43k
LessWrong
[Story] Rejection Just a piece about rejection therapy I did yesterday on the subway. The language might be a bit hyperbolic, but I was kind of on-edge last night. It seemed like it might be a fun story, I want to get better at writing, and gooey self-disclosure seems to be in vogue, so I posted it. If you don't like ...
558b387a-c5f4-4458-bda1-c272e6b6ffd1
trentmkelly/LessWrong-43k
LessWrong
Monthly Linkspost 3/2022 This is a linkspost for https://keller.substack.com/p/monthly-shorts-322 Offense and Defense So, we talk about ransomware as this serious and emerging problem. But it’s interesting to see that part of why ransomware is hard to solve is that it’s incredibly easy. Here are the top criminal gro...
f8d3fa8b-e343-48d7-9909-3034bae35de5
trentmkelly/LessWrong-43k
LessWrong
Superintelligence 23: Coherent extrapolated volition This is part of a weekly reading group on Nick Bostrom's book, Superintelligence. For more information about the group, and an index of posts so far see the announcement post. For the schedule of future topics, see MIRI's reading guide. ----------------------------...
49957a2e-a360-4284-a282-9badfd4b65a2
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Metaphors for AI, and why I don’t like them *[Cross posted on lesswrong  and*[*windowsontheory*](https://windowsontheory.org/2023/06/28/metaphors-for-ai-and-why-i-dont-like-them/)*; see here for my*[*prior writings*](https://windowsontheory.org/category/philosophizing/)*]* “computer, *n. /kəmˈpjuːtə/*. *One who co...
3a5c69f8-61f1-42ab-b63a-c73d13c179e4
StampyAI/alignment-research-dataset/arxiv
Arxiv
Learning to Interactively Learn and Assist 1 Introduction --------------- Many tasks that we would like our agents to perform, such as unloading a dishwasher, straightening a room, or restocking shelves are inherently user-specific, requiring information from the user in order to fully learn all the intricacies of ...
8bb52958-3252-499b-8c56-2d6b0923e2be
StampyAI/alignment-research-dataset/blogs
Blogs
Spurious normativity enhances learning of compliance and enforcement behavior in artificial agents **In** [**our recent paper**](https://www.pnas.org/content/119/3/e2106028118) **we explore how multi-agent deep reinforcement learning can serve as a model of complex social interactions, like the formation of social nor...
86b39677-6c7d-44c1-8262-35f99a18da0b
trentmkelly/LessWrong-43k
LessWrong
Tools to discern between real and AI What are the best ways to figure out if someone who contacted you / a piece of text / an image is AI generated?
5c00eaf4-2eb3-4ff6-b545-2f9f1642222a
trentmkelly/LessWrong-43k
LessWrong
Meetup : Sydney Social Meetup - April (Bridge walk) Discussion article for the meetup : Sydney Social Meetup - April (Bridge walk) WHEN: 10 April 2014 06:30:00PM (+1100) WHERE: Milsons Point Station, Milsons Point NSW 2061, Australia 5:30pm. We will walk across the bridge and eat dinner in the city at around 7ish (...
08c31616-d6e6-4b25-9816-0d16c322a1d9
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post98 In Sparse Feature Circuits ( Marks et al. 2024 ), the authors introduced Spurious Human-Interpretable Feature Trimming (SHIFT), a technique designed to eliminate unwanted features from a model's computational process. They validate SHIFT on the Bias in Bios task, which we think is too simple to serve as mea...
6be599b4-0a3f-40a8-901d-c27c36f1b930
trentmkelly/LessWrong-43k
LessWrong
Yet Another Critique of "Luxury Beliefs" I know little about Rob Henderson except that he wrote a well-received memoir and that he really really really wants you to remember that he invented the concept of “luxury beliefs”. In his own words, these are: > ideas and opinions that confer status on the upper class at ver...
02e934a1-bf25-4112-a63b-b99c34a981a4
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Newcomb's Grandfather In front of you are two boxes. Box A contains $1000. If Omega predicted you would just take Box B, Box B contains $1,000,000. If Omega predicted you would take both boxes, Box B contains nothing. Omega will also go back in time and kill your grandfather. Nobody has any idea ho...
ee6e7cd6-f396-46f8-979e-579a8419903b
trentmkelly/LessWrong-43k
LessWrong
An observation on self-play At NeurIPS 2024, Ilya Sutskever delivered a short keynote address in honor of his Seq2seq paper, published a decade earlier. It was his first—and so far only—public appearance to discuss his research since parting ways with OpenAI. The talk itself shed little light on his current work. Ins...
051a7757-b153-4215-8843-9f89e42b4144
trentmkelly/LessWrong-43k
LessWrong
The Quaker and the Parselmouth
3ef9f2a0-41a6-4818-b552-72528e785e75
trentmkelly/LessWrong-43k
LessWrong
Eliezer on The Lunar Society podcast Another podcast appearance. Great to see Eliezer on a real media blitz tour. > For 4 hours, I tried to come up reasons for why AI might not kill us all, and Eliezer Yudkowsky explained why I was wrong. > We also discuss his call to halt AI, why LLMs make alignment harder, what it...
20fb904a-7b1f-492a-92b7-7c29f92853e8
trentmkelly/LessWrong-43k
LessWrong
Arithmetic is an underrated world-modeling technology Of all the cognitive tools our ancestors left us, what’s best? Society seems to think pretty highly of arithmetic. It’s one of the first things we learn as children. So I think it’s weird that only a tiny percentage of people seem to know how to actually use arithm...
8d74fffb-f7bc-4014-a85e-efd91f9b6aa5
trentmkelly/LessWrong-43k
LessWrong
Avoid misinterpreting your emotions A couple of weeks ago, I was suffering from insomnia. Eventually my inability to fall asleep turned into frustration, which then led to feelings of self-doubt about my life in general. Soon I was wondering about whether I would ever amount to anything, whether any of my various proj...
b21ea3fd-342d-44cb-8d7a-beebc5a93952
trentmkelly/LessWrong-43k
LessWrong
Why Don't Rationalists Win? Here are my thoughts on the "Why don't rationalists win?" thing. Epistemic I think it's pretty clear that rationality helps people do a better job of being... less wrong :D But seriously, I think that rationality does lead to very notable improvements in your ability to have correct beli...
550820a1-97a4-42eb-86d1-b9ab59b34e95
trentmkelly/LessWrong-43k
LessWrong
AI debate: test yourself against chess 'AIs' Hey everyone, @Zane is organising an AI debate around a chess, to test how easy it is for an entity with better understanding to fool one with a lesser understanding. Check out these two threads: Lying to chess players for alignment Deception Chess: Game #1 I thought I wo...
f2810a5d-a0f1-4924-aa5b-f0b65abb59bd
trentmkelly/LessWrong-43k
LessWrong
Meetup : Washington, D.C.: Visiting Museums Discussion article for the meetup : Washington, D.C.: Visiting Museums WHEN: 31 July 2016 03:30:00PM (-0400) WHERE: Donald W. Reynolds Center for American Art and Portraiture This week, we will be meeting in the courtyard to walk together to a museum. As always, side con...
09bac11a-9504-4159-bdec-e864ca5e6675
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Announcing Encultured AI: Building a Video Game *Also available on the* [*EA Forum.*](https://forum.effectivealtruism.org/posts/SbMfbYnMyFYX6acZ2/announcing-encultured-ai-building-a-video-game) *Preceded By:* [Encultured AI Pre-planning, Part 2: Providing a Service](https://www.lesswrong.com/posts/2vxoTfuScspraSJeC...
746f0e49-a20e-4e5e-a5b4-2c5552eaf654
StampyAI/alignment-research-dataset/arxiv
Arxiv
Establishing Appropriate Trust via Critical States I Introduction --------------- When humans have an accurate mental model of a robot, their subsequent interactions with this robot are safer and more seamless. This mental model may include the robot’s intentions [[1](#bib.bib1), [2](#bib.bib2), [3](#bib.bib3)], it...
739fe8b9-e202-4995-8f7e-cbbcaf73ad7a
trentmkelly/LessWrong-43k
LessWrong
Enabling Children There was a recent discussion on Less Wrong about whether or not ambitious people should have kids. I didn't manage to watch it, though I have a lot of interest in the topic. It wasn't recorded but I can still hope for someone to post lots of details about it in the comments. One subtheme from the ...
ce1eb036-3e8b-4d5b-8910-f89b8a8c45ff
trentmkelly/LessWrong-43k
LessWrong
A more effective Elevator Pitch for AI risk " 'We're generating algorithms that we don't really understand, that can do more and more things and keep getting even harder to understand.  They often become capable of doing things we didn't realize they'd be able to do- this is also happening more often.  When we get t...
43d45b74-7289-40d9-aa19-aa078b760adb
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Governing High-Impact AI Systems: Understanding Canada’s Proposed AI Bill. April 15, Carleton University, Ottawa Join us this April 15th for a panel discussion, keynote address by the founder of the Montreal AI Ethics Institute, and networking mixer on the governance of AI Systems in the Canadian context, held by EA C...
054de312-a4ba-4ed7-b5a6-153959601798
trentmkelly/LessWrong-43k
LessWrong
"All natural food" as an constrained optimisation problem A look at all natural foods through the lenses of Bayesianism, optimisation, and friendly utility functions. How should we consider foods that claim to be "all natural"? Or, since that claim is a cheap signal, foods that have few ingredients, all of them easy ...
f014a473-d2d4-438e-9378-a34428b405b2
trentmkelly/LessWrong-43k
LessWrong
Epilogue: Atonement (8/8) (Part 8 of 8 in "Three Worlds Collide") Fire came to Huygens. The star erupted. Stranded ships, filled with children doomed by a second's last delay, still milled around the former Earth transit point.  Too many doomed ships, far too many doomed ships.  They should have left a minute early...
38dc8e91-f2a5-4386-8aed-bdbaf3bc300e
trentmkelly/LessWrong-43k
LessWrong
10 reasons why lists of 10 reasons might be a winning strategy 1. They are really easy to dive into.    2. By making the interesting ideas easy and fun to digest, it is helpful therapy for people who have spent many years with the bad habit of throwing their mind away.   3. It just seems to fit the hum...
d22aed4e-a08e-4365-9520-ac307c15cb9b
trentmkelly/LessWrong-43k
LessWrong
Insights Over Frameworks Once someone has made several insightful self-help observations, I think it's often tempting to then go and find some sort of cohering framework which can explain all of those observations. In this essay, I'll argue for why I think it can be better to skip the post-hoc theorizing altogether an...
4bc2018b-d644-4183-a683-9bc80d759cb3
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Occam's Razor and the Universal Prior > *It can scarcely be denied that the supreme goal of all theory is to make the irreducible basic elements as simple and as few as possible without having to surrender the adequate representation of a single datum of experience. -* [*Albert Einstein*]( https://www.nature.com/artic...
0ca0f155-7e22-451e-a2df-60958a65de69
trentmkelly/LessWrong-43k
LessWrong
DC Meetup: May 22nd Sunday, May 22nd Core hours: 1 PM - 3PM Ballston Common Mall Food Court 4238 Wilson Boulevard Arlington, Virginia 22203 Core hours are 1-3 PM, but many people are willing to hang out longer than that. People have stayed at previous meetups for 4 and 5 hours. We decided to meet in Virginia t...
90f384ed-5cb9-4df5-9d23-edfea4627e26
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
[Meta] Do you want AIS Webinars? Ever now and then I talk to someone who tell me that they can not get good feedback on their research (e.g. the they don't get much responses on their alignment forum post), and been thinking about how to solve this? Also, right now is a good time to try out various online solutions. ...
362e20bd-c950-4917-88c6-72a976d12c7f
trentmkelly/LessWrong-43k
LessWrong
Dynamic Karma & Static Karma ,, The current karma system tracks one number for each user. This number is used both as an indicator of "is this user's writing worth paying attention to" and as a voting weight (using the new powers-of-five system). I think this is hugely broken and splitting these two apart could make t...
f4fe9883-b51a-40eb-adfc-690848af832b
trentmkelly/LessWrong-43k
LessWrong
What do you think of cognitive types and MBTI? What type are you? What do you think is the percentage of the 16 different personality types on LessWrong? None
bf47c87c-5f5c-49ee-bfd6-451fe7b76c22
trentmkelly/LessWrong-43k
LessWrong
Arkose may be closing, but you can help Arkose is an AI safety fieldbuilding organisation that supports experienced machine learning professionals — such as professors and research engineers — to engage with the field. We focus on those new to AI safety, and have strong evidence that our work helps them take meaningfu...
9b81c4d1-deb3-4e09-b106-4ee23eba8dcb
trentmkelly/LessWrong-43k
LessWrong
Liron Shapira vs Ken Stanley on Doom Debates. A review I summarize my learnings and thoughts on Liron Shapira's discussion with Ken Stanley on the Doom Debates podcast. I refer to them as LS and KS respectively. High level summary Key beliefs of KS: * Future superintelligence will be 'open-ended'. Hence, thinking ...
6a3a2b32-5419-4250-a305-f9b77d171b3e
trentmkelly/LessWrong-43k
LessWrong
Meetup : Rationality meetup in Prague #3 Discussion article for the meetup : Rationality meetup in Prague #3 WHEN: 30 June 2015 06:30:00PM (+0200) WHERE: Václavské náměstí 778/14, Praha 110 00 Hello, this is third long-term planed meetup in Prague. We are currently a group of five people regularly meeting every w...
cdcea504-065b-41f0-8698-de9ce69200f2
StampyAI/alignment-research-dataset/blogs
Blogs
MIRI’s May 2014 Newsletter | | | | --- | --- | | | | | --- | | [Machine Intelligence Research Institute](http://intelligence.org) | | | | | | | --- | --- | | | | | --- | | On May 6th, there is $250,000+ in matching funds and prizes available from sources that normally wouldn’t contribute to MIRI at al...
a7932b77-648f-4ceb-8fbe-7b3dc05d25dc
trentmkelly/LessWrong-43k
LessWrong
The Parable Of The Talents [Content note: scrupulosity and self-esteem triggers, IQ, brief discussion of weight and dieting. Not good for growth mindset.] I. I sometimes blog about research into IQ and human intelligence. I think most readers of this blog already know IQ is 50% to 80% heritable, and that it’s so imp...
019ff689-ffd0-49c9-b408-4c817d5e37ce
trentmkelly/LessWrong-43k
LessWrong
[LINK]The Edge's yearly question What is your favorite explanation? http://www.edge.org/responses/what-is-your-favorite-deep-elegant-or-beautiful-explanation And what's yours?   EDIT: I mean yours of those at edge.org - and/or yours which you will explain here.
b54f58b2-8650-47d8-a436-c8a157bdce62
trentmkelly/LessWrong-43k
LessWrong
AI demands unprecedented reliability My AI pessimism is not addressed by optimistic arguments that do not promise the level of reliability required for a technology that our societies vital functions are expected to rely on. Historically, useful improvements are hard to reverse and solutions can suddenly fail after w...
2cd25d27-c696-4e06-ab54-20daa6f07a7d
trentmkelly/LessWrong-43k
LessWrong
g, a Statistical Myth I found this post very interesting It's about statistics, causal inference, and 'g'.
60ddf4ea-1715-40fa-b37b-08bd7c9b7870
trentmkelly/LessWrong-43k
LessWrong
Sparse trinary weighted RNNs as a path to better language model interpretability Epistemic status: Strongly arguing for what I feel is a neglected approach. May somewhat overstate the case and fail to adequately steelman counter arguments. I hope and expect that readers will point out flaws in my logic. Introduction...
9295f4b3-8ed6-49d1-a0a1-8c7aff178cc0
trentmkelly/LessWrong-43k
LessWrong
[Link] Caplan asks for help optimizing his will. Bryan Caplan of Econlog asks his readers how to improve his will (given a few constraints) in light of the principles of optimal philanthropy. His current draft reads: > I give and bequeath to whatever charity is currently ranked #1 by GiveWell, the sum of $100,000 adj...
b9bba649-037a-4df0-ac94-6dc9c4e5a14b
trentmkelly/LessWrong-43k
LessWrong
Acausal trade: being unusual Other posts in the series: Introduction, Double decrease, Pre-existence deals, Full decision algorithms, Breaking acausal trade, Trade in different types of utility functions, Being unusual, and Summary. It's important to remember that, in acausal trade, the other agents truly know nothin...
a18ebf5f-b127-4fea-9bd1-81a04bd7b18e
trentmkelly/LessWrong-43k
LessWrong
Verifiable private execution of machine learning models with Risc0? Risc0 is a virtual machine that runs any riscv (C, C++, Rust, Go, and Nim can all compile to riscv) program, records a trace of the state of the computation over time, and then, no matter how long the execution ran for, it can fold that execution trac...
1ba7f877-5591-4de2-b716-d61b7911f123
trentmkelly/LessWrong-43k
LessWrong
D&D.Sci: Whom Shall You Call? This is a D&D.Sci scenario: a puzzle where players are given a dataset to analyze and an objective to pursue using information from that dataset. Intended Difficulty: Average Good news! After an interminable period of waiting for lawyers to finish arguing over the will, you’ve inherited...
8f9398f6-b464-45e4-85a8-b252a600ab43
trentmkelly/LessWrong-43k
LessWrong
"Go west, young man!" - Preferences in (imperfect) maps Many people are very nationalistic, putting their country above all others. Such people can be hazy about what "above all others" can mean, outside of a few clear examples - eg winning a total war totally. They're also very hazy on what is meant by "their country...
648d0193-6da4-43ed-85fa-1eca29ef5e4a
trentmkelly/LessWrong-43k
LessWrong
Xanadu, GPT, and Beyond: An adventure of the mind Cross posted from New Savanna. I've posted a new working paper. Title above; links, abstract, table of contents, and introductory material below. Download at:  Academia: https://www.academia.edu/106001453/Xanadu_GPT_and_Beyond_An_adventure_of_the_mind  SSRN: https:...
aafe9554-11ee-419c-bde6-224140918016
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
AGI safety from first principles: Alignment *Parts of this section were rewritten in mid-October.* In the previous section, I discussed the plausibility of ML-based agents developing the capability to seek influence for instrumental reasons. This would not be a problem if they do so only in the ways that are aligned...
93554320-1d10-4cae-91f2-7d748d8367e5
trentmkelly/LessWrong-43k
LessWrong
What features would you like a prediction platform to have? None
51a65411-077f-49bb-8524-de29a234e92c
trentmkelly/LessWrong-43k
LessWrong
Self-Awareness: Taxonomy and eval suite proposal [This is a mildly-edited version of a google doc I wrote at OpenAI in July 2022. I had intended to get it published in some form, but never got around to it for various reasons. I have now received approval to put it up as a blog post. The main thing of interest here is...
4ac27c81-080f-4495-b2a5-4f69b20d4ee3
trentmkelly/LessWrong-43k
LessWrong
Babble on growing trust I wonder: "How it happens that people trust someone or something at all?" (in the context of: how can we ever get to trust an AI smarter than us?).  Epistemic status: this is just a dump of my thoughts (inspired by Babble & Prune Thoughts). Feel free to down-vote if you don't agree or don't li...
7948cc4f-527d-4181-ad0e-678337d4c97b
StampyAI/alignment-research-dataset/special_docs
Other
An unaligned benchmark My goal is to design AI systems that are aligned with human interests and [competitive](/directions-and-desiderata-for-ai-control-b60fca0da8f4) with unaligned AI. I find it useful to have a particular AI algorithm in mind. Then I can think about how that algorithm could cause trouble, and try t...
5306cb49-2acb-41f1-8998-7657c5de4ecc
trentmkelly/LessWrong-43k
LessWrong
Book Review: How the World Became Rich In How the World Became Rich, Mark Koyama and Jared Rubin have written what will almost certainly become a beloved economic history classic, as Power and Plenty or The British Industrial Revolution in Global Perspective became over a decade ago. I fawn over a refreshingly readabl...
440cb1d3-9dbf-46f6-a62d-45661dab4ebb
trentmkelly/LessWrong-43k
LessWrong
Rationality Quotes November 2014 Another month, another rationality quotes thread. The rules are: * Please post all quotes separately, so that they can be upvoted or downvoted separately. (If they are strongly related, reply to your own comments. If strongly ordered, then go ahead and post them together.) * Do not...
2df93eb4-ce79-47a8-aeaf-4288cdb0b15a
StampyAI/alignment-research-dataset/blogs
Blogs
A new approach to predicting brain-computer parity *By Katja Grace, 7 May 2015* How large does a computer need to be before it is ‘as powerful’ as the human brain? This is a difficult question, which [people have answered](http://aiimpacts.org/preliminary-prices-for-human-level-hardware/ "Preliminary prices for hu...
91e66e73-7cbd-4f19-89bd-f7cdb318c624
trentmkelly/LessWrong-43k
LessWrong
Covid 3/3/2022: Move Along The number of Ukrainian Covid-19 cases is now ‘nobody gives a f***.’ Quite right. True existential threats put things in perspective. If you are more worried about catching Covid-19 than you are about the war and its possible escalation, you have a poor threat model. That does not mean that...
8d9de729-9d6a-4e32-9b4b-4db803806f0d
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Prizes for ML Safety Benchmark Ideas *“If you cannot measure it, you cannot improve it.” – Lord Kelvin (paraphrased)* Website: [benchmarking.mlsafety.org](https://benchmarking.mlsafety.org/) – receiving submissions until August 2023. ML Safety lacks good benchmarks, so the [Center for AI Safety](https://safe.ai/) is...
2365363e-f08f-487a-9160-1486d78fd8e2
trentmkelly/LessWrong-43k
LessWrong
A case for courage, when speaking of AI danger I think more people should say what they actually believe about AI dangers, loudly and often. Even (and perhaps especially) if you work in AI policy. I’ve been beating this drum for a few years now. I have a whole spiel about how your conversation-partner will react very...
c966f183-9850-404e-8462-35995061c4ba
trentmkelly/LessWrong-43k
LessWrong
Born as the seventh month dies ... Epistemic status: Mathematical reasoning by an amateur, but I feel confident that it’s mostly correct. Update: There is a Wikipedia page on this with lots of details and other similar problems with different answers. This paper referenced there seems a good summary. The problem I ...
2800f1aa-29ec-4160-a17f-9c6a1548f963
trentmkelly/LessWrong-43k
LessWrong
Meetup : Bratislava Discussion article for the meetup : Bratislava WHEN: 01 December 2014 06:00:00PM (+0100) WHERE: Laurinská 1, Bratislava Careful! Now at different place than usual. Discussion article for the meetup : Bratislava
c245efd4-24b6-4af6-badb-9fa2e478277f
trentmkelly/LessWrong-43k
LessWrong
[AN #81]: Universality as a potential solution to conceptual difficulties in intent alignment Find all Alignment Newsletter resources here. In particular, you can sign up, or look through this spreadsheet of all summaries that have ever been in the newsletter. I'm always happy to hear feedback; you can send it to me b...
936e53a5-1af0-4927-9e1b-3cd335c6ee7b
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
The replication and emulation of GPT-3 *This post is one part of the sequence* [*Understanding the diffusion of large language models*](https://forum.effectivealtruism.org/s/8rYkpiFhbb4HsbzFc)*.  As context for this post, I strongly recommend reading at least the* [*5-minute summary*](https://forum.effectivealtruism.o...
e77c066d-8329-4ab7-a197-9f126c84364c
trentmkelly/LessWrong-43k
LessWrong
On the subject of in-house large language models versus implementing frontier models A recent survey in the US states that 39.4% of adults are using generative AI for tasks both at work and outside of work, which highlights the rapidly increasing dependency on these models. Anecdotally, I introduced my wife to chatGPT...