id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
2e6ffd48-cb7a-46e2-9165-7f0ee9064f9d
trentmkelly/LessWrong-43k
LessWrong
Integrated Method for Policy Making Using Argument Modelling and Computer Assisted Text Analysis > IMPACT is conducting original research to develop and integrate formal, computational models of policy and arguments about policy, to facilitate deliberations about policy at a conceptual, language-independent level. The...
bcdec688-cba2-4c81-a7ba-83363395f66b
trentmkelly/LessWrong-43k
LessWrong
Meetup : West LA Meetup 06-28-2011 Discussion article for the meetup : West LA Meetup 06-28-2011 WHEN: 28 June 2011 07:00:00PM (-0700) WHERE: 10800 West Pico Blvd, Suite 312, Los Angeles, CA 90064 When: 7pm - 9pm June 28th. Where: The Westside Pavillion - on the bridge, which connects Nordstrom 3rd floor with Barn...
227d1e25-434a-4520-8eb7-c76c4ab84dcc
trentmkelly/LessWrong-43k
LessWrong
A sketch of an AI control safety case This is a linkpost accompanying a new paper by UK AISI and Redwood Research. Please see the full paper for more details. An illustration of the control evaluation that provides the core evidence used in our control safety case sketch. During the control evaluation, a red team c...
44784cf8-f23d-4636-928c-9cb663d310c5
trentmkelly/LessWrong-43k
LessWrong
Ethical Inhibitions Followup to:  Entangled Truths, Contagious Lies, Evolutionary Psychology What's up with that bizarre emotion we humans have, this sense of ethical caution? One can understand sexual lust, parental care, and even romantic attachment.  The evolutionary psychology of such emotions might be subtler t...
6536711a-6be5-4d68-a251-5ba3af33fb34
StampyAI/alignment-research-dataset/blogs
Blogs
MIRI’s 2018 Fundraiser **Update January 2019**: MIRI’s 2018 fundraiser is now concluded.   [Target 1 $500,000 Completed](https://intelligence.org/feed/?paged=15#fundraiserModal)[Target 2 $1,200,000In Progress](https://intelligence.org/feed/?paged=15#fundraiserModal) **$946,981** | | $0...
6043afdf-e133-43de-8c96-aaafa2671e49
trentmkelly/LessWrong-43k
LessWrong
You'll be who you care about Eliezer wonders about the thread of conscious experience: "I don't think that, if I were really selfish, I could jump off a cliff knowing smugly that a different person would experience the consequence of hitting the ground." Instead of wondering whether we should be selfish towards our f...
e3fbba06-d12e-48e5-8eb8-57b457bfc077
trentmkelly/LessWrong-43k
LessWrong
Is there a definitive intro to punishing non-punishers? Hey LW. I've just written a draft of an article, that I'm trying to edit to link to useful prior notes on various concepts. One of those concepts is the idea that we evolved to "punish the non-punishers", in order to ensure the costs of social punishment are shar...
51d4c3db-b93d-46da-90c2-2f2fb347305d
trentmkelly/LessWrong-43k
LessWrong
And the Winner is... Many-Worlds! This is one of several shortened indices into the Quantum Physics Sequence. Macroscopic quantum superpositions, a.k.a. the "many-worlds interpretation" or MWI, was proposed in 1957 and brought to the general attention of the scientific community in 1970.  Ever since, MWI has steadily...
a6b5ad15-5223-4a2c-bfb8-66e27718d495
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Raised in Technophilia Today's post, Raised in Technophilia was originally published on 17 September 2008. A summary (taken from the LW wiki):   > When Eliezer was quite young, it took him a very long time to get to the point where he was capable of considering that the dangers of technology might outwei...
93c02b8a-eda5-4f94-b727-96196650d658
trentmkelly/LessWrong-43k
LessWrong
Evaluating LLaMA 3 for political sycophancy TLDR: I evaluated LLaMA v3 (8B + 70B) for political sycophancy using one of the two datasets I created. The results for this dataset suggest that sycophancy definitely occurs in a blatant way for both models though more clearly for 8B than for 70B. There are hints of politi...
8c188c63-3653-479d-ad03-9e04f90f09c4
trentmkelly/LessWrong-43k
LessWrong
How can we estimate how many people are C19 infected in an area? My favorite dashboard is Our World in Data: it has time series and maps and lets me choose which countries to graph. But it’s only showing confirmed cases, and doesn’t show any unit smaller than a country. Plus on the day I’m writing this, the US was mi...
c765b1ac-1f05-4955-864c-9a3f3c6612e9
trentmkelly/LessWrong-43k
LessWrong
Does the Universal Geometry of Embeddings paper have big implications for interpretability? Rishi Jha, Collin Zhang, Vitaly Shmatikov and John X. Morris published a new paper last week called Harnessing the Universal Geometry of Embeddings. Abstract of the paper (bold was added by me): > We introduce the first metho...
2aea61a7-4709-48c6-804d-6f33e8a47487
trentmkelly/LessWrong-43k
LessWrong
Market Capitalization is Semantically Invalid Note: I didn't write this essay, nor do I own the blog where it came from. I'm just sharing it to see what other people think about it. The essay text is displayed below this line. In this essay, I will debunk the concept of market capitalization. But first, let’s consid...
417e2e45-bd4d-4a99-8797-15dbba5c7017
trentmkelly/LessWrong-43k
LessWrong
Against Facebook: The Stalking Epistemic Status: Nothing we didn’t basically already know, but felt obligated to share Previously: Against Facebook, Against Facebook: Comparison to Alternatives and Call to Action Also Previously But Fully Optional: Help Us Find Your Blog (and others), How to Destroy Civilization I ...
66ab4958-0f8d-43cb-a4eb-2bc53c5e46f4
trentmkelly/LessWrong-43k
LessWrong
Neural Categories In Disguised Queries, I talked about a classification task of "bleggs" and "rubes".  The typical blegg is blue, egg-shaped, furred, flexible, opaque, glows in the dark, and contains vanadium.  The typical rube is red, cube-shaped, smooth, hard, translucent, unglowing, and contains palladium.  For the...
20b10cca-7fe1-4352-84da-8f7f5e592af4
StampyAI/alignment-research-dataset/distill
Distill Scientific Journal
Self-Organising Textures ### Contents [Patterns, textures and physical processes](#patterns-textures-and-physical-processes) * [From Turing, to Cellular Automata, to Neural Networks](#from-turing-to-cellular-automata-to-neural-networks) * [NCA as pattern generators](#nca-as-pattern-generators) * [Related work](#rela...
c95df073-c2a9-40ff-933d-29b96789425f
trentmkelly/LessWrong-43k
LessWrong
Meetup : Fort Collins, Colorado Meetup Thursday 7pm *New Place* Discussion article for the meetup : Fort Collins, Colorado Meetup Thursday 7pm *New Place* WHEN: 02 August 2012 07:00:00PM (-0600) WHERE: 1129 W. Elizabeth St. Fort Collins 80521 The move to Thursdays is working well. Back to Momo Lomo near campus for ...
672d764f-a79b-4b5d-aa0d-43689f8fcfee
trentmkelly/LessWrong-43k
LessWrong
Is AI safety research less parallelizable than AI research? It seems intuitive to me why that would be the case. And I've seen Eliezer make the claim a few times. But I can't find an article describing the idea. Does anyone have a link?
618d143f-e939-48e0-a752-bb6455ee5077
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post42 We have released a new paper, Ctrl-Z: Controlling AI Agents via Resampling . This is the largest and most intricate study of control techniques to date: that is, techniques that aim to prevent catastrophic failures even if egregiously misaligned AIs attempt to subvert the techniques. We extend control proto...
724059a8-8823-4e53-97ed-62d7d2a022ee
trentmkelly/LessWrong-43k
LessWrong
Holden Karnofsky's Singularity Institute critique: other objections The sheer length of GiveWell co-founder and co-executive director Holden Karnofsky's excellent critique of the Singularity Institute means that it's hard to keep track of the resulting discussion.  I propose to break out each of his objections into a ...
0dd6b93b-f2c7-4205-b000-95aeed1545b4
trentmkelly/LessWrong-43k
LessWrong
ML Safety at NeurIPS & Paradigmatic AI Safety? MLAISU W49 Watch this week’s episode on YouTube or listen to the audio version here. This week, we see how to break ChatGPT, how to integrate diverse opinions in an AI and look at a bunch of the most interesting papers from the ML safety workshop happening right now! C...
b2735c4d-810c-4fc7-919f-9d935ecfc70f
trentmkelly/LessWrong-43k
LessWrong
A toy model of the control problem EDITED based on suggestions for improving the model Jaan Tallinn has suggested creating a toy model of the control problem, so that it can be analysed without loaded concepts like "autonomy", "consciousness", or "intentionality". Here a simple (too simple?) attempt:   A controls ...
d2695f0f-97f2-4fe6-ab3c-895989ae2b32
trentmkelly/LessWrong-43k
LessWrong
Non-Monotonic Infra-Bayesian Physicalism Infra-Bayesian physicalism (IBP) is a mathematical formalization of computationalist metaphysics: the view that whether something exists is a matter of whether a particular computation is instantiated anywhere in the universe in any way. In this post, we cover the basics of IBP...
d6b50954-3b31-4c65-a67a-205a31516647
trentmkelly/LessWrong-43k
LessWrong
My Reservations about Discovering Latent Knowledge (Burns, Ye, et al) [This is a cross-post from my blog at aizi.substack.com.] This is my second post on Burns, Ye, et al’s recent preprint Discovering Latent Knowledge in Language Models Without Supervision. My first post summarizes what they did and what I liked abou...
d7e91aca-0694-493d-bdc7-578dc8cc8a96
trentmkelly/LessWrong-43k
LessWrong
Necessary, But Not Sufficient There seems to be something odd about how people reason in relation to themselves, compared to the way they examine problems in other domains. In mechanical domains, we seem to have little problem with the idea that things can be "necessary, but not sufficient".  For example, if your car...
7eeb668c-0526-4f44-b260-2855504f3ecf
trentmkelly/LessWrong-43k
LessWrong
Existential AI Safety is NOT separate from near-term applications Stephen Casper, scasper@mit.edu. Thanks to Alex Lintz and Daniel Dewey for feedback.  This is a reply but not an objection to a recent post from Paul Christiano titled AI alignment is distinct from its near term applications. The post is fairly brief a...
81dd25cf-d7af-48f8-990f-6e0075fdf668
trentmkelly/LessWrong-43k
LessWrong
How far along are you on the Lesswrong Path? How far along are you on the Path? Some say the Path has always existed and there have always been those who have walked it. The Path has many names and there are many notions of what the Path is. To some the path is called philosophy, to others it is the art of rationalit...
7275a076-e196-4d33-9634-71cc6e1724e7
trentmkelly/LessWrong-43k
LessWrong
Sensual Experience Modern day gamemakers are constantly working on higher-resolution, more realistic graphics; more immersive sounds—but they're a long long way off real life. Pressing the "W" key to run forward as a graphic of a hungry tiger bounds behind you, just doesn't seem quite as sensual as running franticall...
991041ff-0c5e-4b7a-80af-6c062f5416c0
trentmkelly/LessWrong-43k
LessWrong
Strategyproof Mechanisms: Impossibilities In which the limits of dominant-strategy implementation are explored. The Gibbard-Satterthwaite dictatorship theorem for unrestricted preference domains is stated, showing no universal strategyproof mechanisms exist, along with a proof for a special case. Due to the Revelatio...
9b723294-abde-4e44-aa33-f941a48b2a77
trentmkelly/LessWrong-43k
LessWrong
Weak evidence that eating vegetables makes you live longer People vary in how much they can taste bitter things. If you go around giving people the chemical Phenylthiocarbamide (PTC), which you shouldn't do, because it is toxic, you'll find that some people taste it as a strongly bitter while others can't taste it at ...
83a3a388-252a-44f0-a7a7-20a3174f6c85
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
2-D Robustness *This is a short note on a framing that was developed in collaboration with Joar Skalse, Chris van Merwijk and Evan Hubinger while working on [Risks from Learned Optimization](https://arxiv.org/abs/1906.01820), but which did not find a natural place in the report.* --- Mesa-optimisation is a kind of...
a29802c2-8893-424e-b9a1-8d6aec28b6ce
trentmkelly/LessWrong-43k
LessWrong
Proving the Geometric Utilitarian Theorem This is a supplemental post to Geometric Utilitarianism (And Why It Matters), which sets out to prove what I think are the main interesting results about Geometric Utilitarianism: * Maximizing a geometric weighted average G(_,ψ) can always lead to Pareto optimality. * Given...
3b15bc94-df10-43fa-b693-45b25b0c01d9
trentmkelly/LessWrong-43k
LessWrong
The Real Fanfic Is The Friends We Made Along The Way [This is a humorous short story about an AI that is almost aligned... but not quite. It was originally written for the 2022 EA writing retreat, Ink At The Abbey, and was generally well received] The Real Fanfic Is The Friends We Made Along The Way by Eneasz Brodsk...
9653549d-205b-4807-8144-a1e4bd5d46fa
trentmkelly/LessWrong-43k
LessWrong
Desiderata for Normalizers This work originated at MIRI Summer Fellows and originally involved Pasha Kamyshev, Dan Keys, Johnathan Lee, Anna Salamon, Girish Sastry, and Zachary Vance. I was asked to look over two drafts and some notes, clean them up, and post here. Especial thanks to Zak and Pasha for drafts on which ...
50ad8086-14ff-4615-bd61-7fe25353b2e0
trentmkelly/LessWrong-43k
LessWrong
Luck based medicine: angry eldritch sugar gods edition Introduction Epistemic status: everything is stupid. I’m pretty sure I’m directionally right but this post is in large part correcting previous statements of mine, and there’s no reason to believe this is the last correction. Even if I am right, metabolism is high...
0c30227b-2d25-45e2-b24d-a892f55c0efb
trentmkelly/LessWrong-43k
LessWrong
What we Know vs. How we Know it? Two weeks ago I said: > The other concept I’m playing with is that “what we know” is inextricable from “how we know it”. This is dangerously close to logical positivism, which I disagree with my limited understanding of. And yet it’s really improved my thinking when doing historical r...
19265666-70bc-41b1-bce6-1ffbdf25be2f
StampyAI/alignment-research-dataset/special_docs
Other
Fixed-point solutions to the regress problem in normative uncertainty Introduction People sometimes make claims about how we ought to act in the face of empirical uncertainty. A "decision theory" is a collection of such claims. Because they make demands on our behavior, decision theories are "norms". (Descriptive clai...
c297854f-2cb9-43ec-8bcd-7702148cf18f
trentmkelly/LessWrong-43k
LessWrong
Frequent sauna bathing protects men against dementia
5de763fb-c475-4ef2-8078-793b667d1089
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
SIA is basically just Bayesian updating on existence I love the [sleeping Beauty problem](https://en.wikipedia.org/wiki/Sleeping_Beauty_problem), but I think it leads people astray, by making them think that anthropic reasoning is about complicated situations with multiple duplicates. That leads to people saying erron...
79887ab2-cf3a-44bf-9304-92f66db489f9
trentmkelly/LessWrong-43k
LessWrong
Anthropic CEO calls for RSI "If China can't get millions of chips, we'll (at least temporarily) live in a unipolar world, where only the US and its allies have these models. It's unclear whether the unipolar world will last, but there's at least the possibility that, because AI systems can eventually help make even s...
8c5ff644-be45-4fae-8a5f-f8e0037d6f6a
trentmkelly/LessWrong-43k
LessWrong
To Match the Greats, Don’t Follow In Their Footsteps People who dream of being like the great innovators in history often try working in the same fields - physics for people who dream of being like Einstein, biology for people who dream of being like Darwin, etc. But this seems backwards to me. To be as revolutionary...
75f1ccd9-2e37-40ab-91e4-571da5369d85
trentmkelly/LessWrong-43k
LessWrong
The Cryonics Strategy Space In four paragraphs I’m going to claim, “It is highly likely reading this article will increase your chance of living forever”. I’m pretty sure you won’t disagree with me. First, however, I’d like to talk about how much I don’t like Monopoly. I play a lot of Monopoly, because I am forced in...
e2a52c66-d255-4f40-98e9-f042c88ef97b
trentmkelly/LessWrong-43k
LessWrong
What If: An Earthquake in Taiwan? Briefly and without any research myself, I learned today that Taiwan, where most of the most advanced chip fabs are, is very prone to earthquakes. Which are bad for very-high-precision manufacturing. It would also be right in the middle of any conflict between the US and China. How ...
5a09ac20-338a-4d5b-8d0e-6d7781c87b09
trentmkelly/LessWrong-43k
LessWrong
What is your Personal Knowledge Management system? > Personal knowledge management (PKM) is a collection of processes that a person uses to gather, classify, store, search, retrieve and share knowledge in their daily activities (Grundspenkis 2007) and the way in which these processes support work activities (Wright 20...
9b28460a-1bcb-4f8d-b009-554d8354ee87
trentmkelly/LessWrong-43k
LessWrong
Consider working more hours and taking more stimulants Epistemic status: amphetamines > Sooner or later, the great men turn out to be all alike. They never stop working. They never lose a minute. It is very depressing. > > — V.S. Pritchett EA produces much talk about an obligation to donate some or even most of you...
1f8ef4b5-cf56-4e73-81c4-4c680b42cb85
trentmkelly/LessWrong-43k
LessWrong
Beyond Democracy: A System Where Citizens Vote with Their Taxes A problem with all systems of government is that whenever a person is given power over something, he can do whatever he wants with that power. An example might be that citizens vote to fund a police force, but then the police extract protection money from...
3c909d47-2c86-47c0-bc9a-85656b68a7b6
StampyAI/alignment-research-dataset/arxiv
Arxiv
Towards the Quantification of Safety Risks in Deep Neural Networks 1 Introduction --------------- In recent years, we witness significant progress has been made in AI, especially the deep neural networks that can achieve surprisingly high performance on various tasks, including image recognition [[1](#bib.bib1)], n...
7316a01b-a0fa-45c4-afb8-2540340315e9
trentmkelly/LessWrong-43k
LessWrong
Lies Told To Children Growing up, as a kid, I was always told that every sapient life is precious, everything that thinks and knows itself - Yes, this is a tale about lies-told-to-children.  You'll probably figure it out yourself before too long.  For now, just listen. Where was I?  Right.  As children, we were alwa...
d40e2450-3839-41af-bb37-205651725afa
trentmkelly/LessWrong-43k
LessWrong
Educating people about rationality: where are we? Motivation. There are a number of ideas in rationality that seem to me so important and useful that it's worthwhile systematically exposing the general population to them. Off the top of my head, these might include general ideas about better decision making, cognitive...
4643991d-34a1-44d7-b7f9-1b1b238b9762
trentmkelly/LessWrong-43k
LessWrong
Does turning on the shower help reduce wildfire smoke in the air? Rain is said to help air quality not only by stopping fires but also by removing smoke particles from the air. Does turning on the shower also remove smoke particles from the air, or does something different happen higher up in the atmosphere vs in a sh...
11bdce13-51ab-4608-8546-8e4d86cdc47b
trentmkelly/LessWrong-43k
LessWrong
Reaction and Reply to Sasha Chapin on Bad In-group Norms EPISTEMIC STATUS: STREAM-OF-CONSCIOUSNESS Earlier today, my co-blogger Devin posted Sasha Chapin's article, "Your Intelligent, Conscientious In-group Has Bad Social Norms Too" to a mutual Discord server. He specifically wanted to know our friend group's thought...
a14a095b-25f2-4103-bae1-ec058855051d
trentmkelly/LessWrong-43k
LessWrong
Decision Theory is multifaceted Related: Conceptual Problems with UDT and Policy Selection, Formalising decision theory is hard Target Anyone who is interested in decision theory. The post is pretty general and not really technical; some familiarity with counterfactual mugging can be useful, but overall the required...
0ef62324-5f86-4b13-bb15-d0f04796f4e0
StampyAI/alignment-research-dataset/arbital
Arbital
Intradependent encoding We call an [https://arbital.com/p/encoding](https://arbital.com/p/encoding) $E(m)$ of a message $m$ "intradependent" if the fact that $E(m)$ encodes $m$ can be deduced without looking at the whole encoding. For example, imagine that you know you're going to see an 8-letter English word, and you...
eae9d154-c2af-4dca-8783-b0307fab5c80
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Rant on Problem Factorization for Alignment This post is the second in what is likely to become a series of uncharitable rants about alignment proposals (previously: [Godzilla Strategies](https://www.lesswrong.com/posts/DwqgLXn5qYC7GqExF/godzilla-strategies)). In general, these posts are intended to convey my underlyi...
eabab0ec-71fe-4da5-af8c-87e197e88c0d
trentmkelly/LessWrong-43k
LessWrong
Subscripting Typographic Convention For Citations/Dates/Sources/Evidentials: A Proposal > Reviving an old General Semantics proposal: borrowing from scientific notation and using subscripts like 'Gwern2020' for denoting sources (like citation, timing, or medium) might be a useful trick for clearer writing, compared to...
6b5a78fe-2130-4d96-952d-c9d07138c4e9
trentmkelly/LessWrong-43k
LessWrong
Reflective Consequentialism Epistemic Status: My position makes sense to me and seems pretty plausible. However, I haven't thought too hard about it, nor have I put it to the test by discussing with others. Putting it to the test is what I'm trying to do here. I also don't have the best understanding of what virtue et...
96142dba-7bef-44c1-bbb9-8f0ac5107f67
trentmkelly/LessWrong-43k
LessWrong
Deadlines and AI theory The AI is a real-time algorithm - it has to respond to situation in the real time. The real-time systems have to trade time for accuracy, and/or face deadlines. The straightforward utility maximization may look viable for multiple choice questions, but for write-in problems, such as technologi...
1fc2248e-42a5-415a-be99-6d5df24b0c3a
trentmkelly/LessWrong-43k
LessWrong
Samotsvety's AI risk forecasts Crossposted to EA Forum and Foxy Scout Introduction In my review of What We Owe The Future (WWOTF), I wrote: > Finally, I’ve updated some based on my experience with Samotsvety forecasters when discussing AI risk… When we discussed the report on power-seeking AI, I expected tons of sk...
7987ee34-a9f8-455c-adb8-d7d9926bccd9
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
It Looks Like You're Trying To Take Over The World *This story was originally posted as a response to* [*this thread*](https://forum.effectivealtruism.org/posts/DuPEzGJ5oscqxD5oh/shah-and-yudkowsky-on-alignment-failures?commentId=hjd7Z4AN6ToN2ebSm)*.* *It might help to imagine a hard takeoff scenario using only know...
8eb7af70-77ad-482c-8453-245e7a194c66
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Decoupling deliberation from competition I view [intent alignment](https://ai-alignment.com/clarifying-ai-alignment-cec47cd69dd6) as one step towards a broader goal of decoupling deliberation from competition*.* * **Deliberation**. Thinking about what we want, learning about the world, talking and learning from each ...
408e724e-3ba4-4a85-aa7d-d59789b884e7
trentmkelly/LessWrong-43k
LessWrong
Cambridge LW: Post-mortem   A brief post-mortem of this weeks LW meet-up. Topic Selection This week we had comparatively few people, so what worked this time may not generalise. At the beginning of the meet-up Douglas suggested we use one of his meetup formats. Of the three, skill focus was not usable without prepa...
5552144e-2c49-4395-99d5-4fddd63c6cab
StampyAI/alignment-research-dataset/arxiv
Arxiv
Playing hard exploration games by watching YouTube 1 Introduction --------------- People learn many tasks, from knitting to dancing to playing games, by watching videos online. They demonstrate a remarkable ability to transfer knowledge from the online demonstrations to the task at hand, despite huge gaps in timing...
3d13b39a-d919-4856-922c-5d1ac268f5e8
trentmkelly/LessWrong-43k
LessWrong
Official LW uncensored thread (on Reddit) http://www.reddit.com/r/LessWrong/comments/17y819/lw_uncensored_thread/ This is meant as an open discussion thread someplace where I won't censor anything (and in fact can't censor anything, since I don't have mod permissions on this subreddit), in a location where comments a...
09fb85e1-77b0-4d22-8482-3239b52d0cc5
StampyAI/alignment-research-dataset/blogs
Blogs
Coherence arguments imply a force for goal-directed behavior **By Katja Grace, 25 March 2021** *[Epistemic status: my current view, but I haven’t read all the stuff on this topic even in the LessWrong community, let alone more broadly.]* There is a line of thought that says that advanced AI will tend to be ‘goal-d...
ef6d8e4e-5455-4a7a-a5dc-e4c1533089ac
trentmkelly/LessWrong-43k
LessWrong
... Wait, our models of semantics should inform fluid mechanics?!? This post is styled after conversations we’ve had in the course of our research, put together in a way that hopefully highlights a bunch of relatively recent and (ironically) hard-to-articulate ideas around natural abstractions.  John: So we’ve been w...
173deda2-9719-4bb4-ae12-4234b59a06e0
trentmkelly/LessWrong-43k
LessWrong
I universally trying to reject the Mind Projection Fallacy—consequences Jan 2025 - translation updated My essay is a continuation of the reasoning from Yudkowsky's article "Mind Projection Fallacy", and an attempt to expand it by providing more gears of how the fallacy works. When I understood the mechanism described...
012b052f-09f7-48df-8d42-7030e90c3024
trentmkelly/LessWrong-43k
LessWrong
What does the launch of x.ai mean for AI Safety? x.ai has now launched. It seems worthwhile to discuss both what it means for AI safety and whether people interested in AI safety should consider applying for the company. Some thoughts: It's notable that Dan Hendrycks is listed as an advisor (the only advisor listed)...
ecacc43d-1f59-49f8-93fb-2655f39a6d9d
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] AI Go Foom Today's post, AI Go Foom was originally published on November 19, 2008. A summary:   > Robin Hanson's attempt to summarize Eliezer Yudkowsky's position. Discuss the post here (rather than in the comments to the original post). This post is part of the Rerunning the Sequences series, where w...
9454bef8-5eca-47bc-a1f0-8b62558ce72b
trentmkelly/LessWrong-43k
LessWrong
Knowledge Base 9: Increasing intelligence The post Knowledge Base 2: The structure and the method of building I proposed to build a database of credible knowledge, understandable both to people and computers, including AI. This database could be used as a source of information in computer applications, and to increase...
5719ce2c-8f2d-4b05-a870-28ac97b8fff6
trentmkelly/LessWrong-43k
LessWrong
Superrationality and network flow control Computers exchanging messages on a network must decide how fast or slow to transmit messages. If everyone transmits too slowly then the network is underutilized, which is bad for all. If everyone transmits too quickly then most messages on the network are actually flow control...
b5fcf736-fdb9-402a-b328-b4054a6f28d0
trentmkelly/LessWrong-43k
LessWrong
Minneapolis Meetup, This Saturday (26th) at 3:00 PM, University of Minnesota The Minneapolis meetup will take place this Saturday the 26th at 3:00 PM in Coffman Memorial Union at the University of Minnesota. Directions and address can be found here, and we'll be meeting inside near Starbucks, next to the northwest en...
dad7c17d-a405-4eeb-9054-8896974b2bf0
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
Alex Turner - Will powerful AIs tend to seek power? [Music] hey everyone welcome back to the towards data science podcast now today's episode is somewhat special because we're going to be talking about what might be the first solid quantitative study of the power seeking tendencies that we can expect advanced ai syste...
191ffb6d-29dd-4ba9-8236-0a77e89be484
trentmkelly/LessWrong-43k
LessWrong
Emergency Residential Solar Jury-Rigging With generous incentives, many houses around here now have rooftop solar. In an emergency, however, these panels would not be able to provide electricity: we've regrettably chosen to set things up where almost all residential solar only provides power when the grid is up. But ...
b758552f-0479-49dd-8cdf-7616d535826c
trentmkelly/LessWrong-43k
LessWrong
Engineering archaeology Here's an account by a retired engineer of what happened when his old company wanted to streamline a process in the factory where he used to work. People only knew how to keep the factory going from one day to the next, but all the documentation was lost-- the factory had been sold a couple o...
e2477102-15e1-42ac-8cd0-b247d58a455f
trentmkelly/LessWrong-43k
LessWrong
Glide #1.5: New Criticism and Rationality Glide Meditation #1.5: A Reflection on Reflecting on the Original Sequences When I started to write (what I anticipated to be) my second Glide Meditation, I almost immediately hit a wall. Now that I had something concrete to work with, I couldn't simply smile, cheerily quip "...
848f8140-8c37-45b7-98df-8eef4160be07
trentmkelly/LessWrong-43k
LessWrong
The Learning-Theoretic Agenda: Status 2023 TLDR: I give an overview of (i) the key problems that the learning-theoretic AI alignment research agenda is trying to solve, (ii) the insights we have so far about these problems and (iii) the research directions I know for attacking these problems. I also describe "Physical...
25424114-bd30-48cc-860e-cb1509d8642a
trentmkelly/LessWrong-43k
LessWrong
Relativity Theory for What the Future 'You' Is and Isn't "Me" encompasses three constituents: this mind here and now, its memory, and its cared-for future. There follows no ‘ought’ with regards to caring about future clones or uploadees, and your lingering questions about them dissipate. In When is a mind me?, Rob Be...
a3403ec5-6cb8-423e-b3b5-5374e35f59bb
trentmkelly/LessWrong-43k
LessWrong
Mini advent calendar of Xrisks: nanotechnology The FHI's mini advent calendar: counting down through the big five existential risks. The third one is a also a novel risk: nanotechnology.   Nanotechnology Current understanding: low Most worrying aspect: the good stuff and the bad stuff are the same thing The potent...
c3eb0b53-e5c4-4a82-a8a0-b55994ff3e48
trentmkelly/LessWrong-43k
LessWrong
Meetup : Berkeley meetup: Board game night Discussion article for the meetup : Berkeley meetup: Board game night WHEN: 01 August 2012 07:00:00PM (-0700) WHERE: Berkeley, CA Hello, all! Tomorrow's meetup will be a board game night at Zendo. Zendo's game library has Robo Rally, Smallworld, Dominion, Ticket to Ride, S...
9b250ec5-4bbc-4869-9b39-c98391bf134d
trentmkelly/LessWrong-43k
LessWrong
Neil deGrasse Tyson on Cryonics Question: > What are your thoughts on cryogenic preservation and the idea of medically treating aging? His response: > A marvelous way to just convince people to give you money. Offer to freeze them for later. I'd have more confidence if we had previously managed to pull this off wit...
5ae66071-e3b7-48b9-ad4f-98f4547ccff1
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"When I lecture on the Singularity, I often draw a graph of the "scale of intelligence" as it appears in everyday life: But this is a rather parochial view of intelligence. Sure, in everyday life, we only deal socially with other humans—only other humans are partners in the great game—and so we only meet the minds of i...
15c2c88c-3735-4b38-ba23-0c42786f7582
trentmkelly/LessWrong-43k
LessWrong
Victoria Krakovna on AGI Ruin, The Sharp Left Turn and Paradigms of AI Alignment Victoria Krakovna is a Research Scientist at DeepMind working on AGI safety and a co-founder of the Future of Life Institute, a non-profit organization working to mitigate technological risks to humanity and increase the chances of a posi...
e1ce7d34-ac20-4783-b566-bc893f9ec390
trentmkelly/LessWrong-43k
LessWrong
(Not) Derailing the LessOnline Puzzle Hunt (spoiler alert: may meta-spoil future iterations of the LOPH, if you haven’t already read other posts about it) I knew early on that I wouldn’t be able to finish the LessOnline puzzle hunt. I contributed to solving two of the first six puzzles, each of which revealed the co...
6f985365-d3f2-436e-b5b7-6d1e519edb1d
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Victoria Krakovna on AGI Ruin, The Sharp Left Turn and Paradigms of AI Alignment Victoria Krakovna is a Research Scientist at DeepMind working on [AGI safety](https://deepmindsafetyresearch.medium.com/) and a co-founder of the [Future of Life Institute](https://futureoflife.org/), a non-profit organization working to ...
3c1190ab-c91c-4c1c-bce4-c0aec71a6962
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Metaculus Predicts Weak AGI in 2 Years and AGI in 10 Just a quick update on predicted timelines. Obviously, there’s no guarantee that Metaculus is 100% reliable + you should look at other sources as well, but I find this concerning. Weak AGI is now predicted in a little over two years: <https://www.metaculus.com/q...
298923f8-456f-40a8-996d-3ebf0ba45c6e
trentmkelly/LessWrong-43k
LessWrong
Clarifying Power-Seeking and Instrumental Convergence Previously: Seeking Power is Provably Instrumentally Convergent in MDPs Rohin Shah and Vanessa Kosoy pointed out a subtle problem with my interpretation of the power-seeking theorem from the last post. To understand the distinction, we first need to shore up some ...
4b6a7508-3ea8-4c36-a73e-ca60dea2310c
trentmkelly/LessWrong-43k
LessWrong
The poetry of progress I’ve said that Western culture, before the World Wars, had a positive philosophy of progress. How do we know that people in that era saw glory, romance, even poetry in progress? One way we know is that they literally wrote poetry about it. Take, for example, Erasmus Darwin (grandfather of Charl...
84697dd9-6fbc-4d37-a5c3-a05ad17e6828
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Announcing The Most Important Century Writing Prize **tl;dr:  Win**[**$500 prizes**](https://www.super-linear.org/prize?recordId=recmsHjuOh5JmVJWJ)**for writing posts that expand on Holden Karnofsky's**[**Most Important Century series.**](https://www.cold-takes.com/most-important-century/) Introducing the prize -----...
d60bade2-4ad4-4eee-9f12-987cd06117ef
StampyAI/alignment-research-dataset/lesswrong
LessWrong
ARC tests to see if GPT-4 can escape human control; GPT-4 failed to do so An unorthodox approach to alignment research to be sure. The advice to "fail early" might actually a good thing when it comes to AI x-risk though. After this post was written, ARC themselves wrote a much more detailed (and perhaps more accurate...
4952d98b-2ff2-45a1-8035-b179d1b53349
trentmkelly/LessWrong-43k
LessWrong
Meetup : Cambridge, MA third-Sundays meetup Discussion article for the meetup : Cambridge, MA third-Sundays meetup WHEN: 18 November 2012 02:00:00PM (-0500) WHERE: 25 Ames St Cambridge, MA Cambridge/Boston-area Less Wrong meetups on the first and third Sunday of every month, 2pm at the MIT Landau Building [25 Ames ...
65a9eb3b-8a58-44eb-b8e8-d36b262fe213
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Safer sandboxing via collective separation *Epistemic status: speculative brainstorming. Follow-up to [this post on AGIs as collectives](https://www.alignmentforum.org/posts/HekjhtWesBWTQW5eF/agis-as-populations). Note that I've changed the term* population AGI *to* collective AGI *for consistency with Bostrom's use i...
7b132ac8-008f-4f78-932a-646ada35d881
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
A distillation of Evan Hubinger's training stories (for SERI MATS) *This post is a* [*distillation*](https://www.lesswrong.com/posts/zo9zKcz47JxDErFzQ/call-for-distillers) *of Evan Hubinger's post "*[*how do we become confident in the safety of a machine learning system?*](https://www.alignmentforum.org/posts/FDJnZt8K...
753d824e-3a1e-455c-91ac-c86147bfc71d
StampyAI/alignment-research-dataset/arbital
Arbital
Free group Intuitively, the free [https://arbital.com/p/-3gd](https://arbital.com/p/-3gd) $F(X)$ on the [set](https://arbital.com/p/3jz) $X$ is the group whose elements are the [freely reduced words](https://arbital.com/p/5jc) over $X$, and whose group operation is "[https://arbital.com/p/-concatenation](https://arbit...
934edbf4-51c5-4cb4-9289-05515ea02717
StampyAI/alignment-research-dataset/lesswrong
LessWrong
How much should we worry about mesa-optimization challenges? *This post was written quickly, lest I do not write it at all.* Picture the following scenario.  1. Humans train a model, M, with the intention for M to minimize a loss function L. 2. The model, M, will now take a set of actions. I see this going wrong in...
bd394baf-8bf0-4e4a-933c-bba6ab8d1450
trentmkelly/LessWrong-43k
LessWrong
Neuroscience and Alignment I've been in many conversations where I've mentioned the idea of using neuroscience for outer alignment, and the people who I'm talking to usually seem pretty confused about why I would want to do that. Well, I'm confused about why one wouldn't want to do that, and in this post I explain why...
0ef5dd3c-083c-458c-986b-791669287a73
trentmkelly/LessWrong-43k
LessWrong
Meetup : Bielefeld Meetup, February 20th Discussion article for the meetup : Bielefeld Meetup, February 20th WHEN: 20 February 2013 07:00:00PM (+0100) WHERE: Grill/Bar Verve, Klosterplatz 13, Bielefeld We are meeting once again in Bielefeld. The topics of this evening are not yet determined, but will be in the ne...
b590b4d2-eac5-460b-b211-4146abbe0b8d
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"This is an exercise, and as such is a bit different from your ordinary question post...What? Come up with 50 ways of sending something to the moon. In less than 1 hour. I don’t care how stupid they are. My own list included “Slingshot”, “Massive trampoline” and “Bird with spacesuit”.What matters is that you actually h...
13d9ae0c-913e-48c1-a558-9b7582be743d
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"There’s a programming technique where you keep a rubber duck on your desk, and when you run into a confusing bug, you try to explain it to the duck. This turns out to be surprisingly useful for resolving bugs.In research conversations, it is often very useful to have someone serve as an upgraded rubber duck. This is a...
e15d4d93-e4aa-40e4-8bb6-de953bf73255
trentmkelly/LessWrong-43k
LessWrong
How do we build organisations that want to build safe AI? Originally posted here. Much has been written about the dangers of artificial intelligence, and how the intelligent systems we build may sometimes unintentionally drift in alignment from their original goals. However, we seem to focus almost entirely upon an A...
99c93c5e-579d-483a-ab9a-486ec6c2db87
trentmkelly/LessWrong-43k
LessWrong
The Alignment Trap: AI Safety as Path to Power Recent discussions about artificial intelligence safety have focused heavily on ensuring AI systems remain under human control. While this goal seems laudable on its surface, we should carefully examine whether some proposed safety measures could paradoxically enable rath...