id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
22991b3c-4e2f-476a-a108-a00ee4e6fb20
trentmkelly/LessWrong-43k
LessWrong
Georgism, in theory Last year, Astral Codex had a series of interesting posts on Georgism. The claims felt very counterintuitive to me, so I dove into it until I could at least sort out the theory. And I emerged with a better understanding, which I think would be useful to share here. Key points: 1. In theory at le...
e73548a7-c57b-4cec-bdd7-b5d87fcebafd
trentmkelly/LessWrong-43k
LessWrong
The Other Alignment Problem: Maybe AI Needs Protection From Us Imagine you're building an AI. You spend enormous amounts of time and resources carefully ensuring it will never hurt humans. Your entire field—the "AI alignment" community—is dedicated to protecting humanity from an AI apocalypse. But what if we've got th...
349b322b-f1dd-4173-a0f7-475bccb47a35
trentmkelly/LessWrong-43k
LessWrong
Jailbreaking ChatGPT on Release Day ChatGPT is a lot of things. It is by all accounts quite powerful, especially with engineering questions. It does many things well, such as engineering prompts or stylistic requests. Some other things, not so much. Twitter is of course full of examples of things it does both well and...
1bf51da1-f1c2-447c-952e-5f1b5544d5c4
trentmkelly/LessWrong-43k
LessWrong
Meta: What tool turns rich text into clean HTML? If you write an article in Word, Writer, Scrivener, Google Docs, or another rich text editor, and then copy+paste that rich text into an online WYSIWYG editor like the one on Less Wrong or WordPress, the HTML generated by LW or WordPress is incredibly messy and does ton...
fbb7f8fe-197d-4c8c-a02f-876518b39ae7
trentmkelly/LessWrong-43k
LessWrong
What empirical work has been done that bears on the 'freebit picture' of free will? This picture was described in Scott Aaronson's essay The Ghost in the Quantum Turing Machine in 2013, and claims that human free will is related to our choices being caused by (and/or causing) quantum bits from the initial state of the...
611fb1cd-1058-45f3-b719-8102d4f10259
trentmkelly/LessWrong-43k
LessWrong
Upcoming AI regulations are likely to make for an unsafer world TL;DR: I am rather confident that involving governments in regulating AI development will make the world less safe, not more, because governments are basically incompetent and the representatives are beholden to special interests. Governments can do very...
68dc89cc-e2d3-499c-b4b2-fb32d638889e
trentmkelly/LessWrong-43k
LessWrong
Cowering To Genocide: Uighur Persecution And The World’s Last Hope If you could stop a genocide-would you?  In western China, right now, more than a million men, women, and children sit in reeducation camps. Mothers separated from children, forced marriages, forced sterilizations-Margaret Atwood’s fiction made a real...
a254156c-a97d-4b66-b1ba-a7fe49b144ce
StampyAI/alignment-research-dataset/special_docs
Other
X2T: Training an X-to-Text Typing Interface with Online Learning from User Feedback. Published as a conference paper at ICLR 2021 X2T: T RAINING AN X-TO-TEXT TYPING INTERFACE WITH ONLINE LEARNING FROM USERFEEDBACK Jensen Gao, Siddharth Reddy, Glen Berseth University of California, Berkeley sgr@berkeley.eduNicholas Har...
7ee23b9d-5f27-4833-807e-3b4c19e342ad
trentmkelly/LessWrong-43k
LessWrong
Some disjunctive reasons for urgency on AI risk (This has been sitting in my drafts folder since August 2017. Robin Hanson's recent How Lumpy AI Services? made me think of it again. I'm not sure why I didn't post it back then. I may have wanted to add more reasons, details and/or citations, but at this point it seems ...
71940ac3-938f-41a3-9bd9-cee3af80dadf
trentmkelly/LessWrong-43k
LessWrong
What is "Instrumental Corrigibility"? In his article here, Paul Christiano mentions the idea that instrumentally corrigible agents are not robust, as slight misalignments between their values and the overseer values would lead to an adversarial relationship. While the intuition that slight deviations in values potenti...
4fedc38b-cf09-4505-8113-cdc3a6c67af4
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
A dataset for AI/superintelligence stories and other media? ([Ahem](https://forum.effectivealtruism.org/posts/6whiBq7czKJk4Bx29/a-forum-post-can-be-short)) I'm curious whether: 1. Anyone is aware of a database/spreadsheet of popular media related to AI, including works such as the [Terminator](https://forum.effectiv...
78397b8b-3145-45c5-bc8a-9ac33685bee9
trentmkelly/LessWrong-43k
LessWrong
(Appetitive, Consummatory) ≈ (RL, reflex) “Appetitive” and “Consummatory” are terms used in the animal behavior literature. I was was briefly confused when I first came across these terms (a year or two ago), because I’m most comfortable thinking in terms of brain algorithms, whereas these terms were about categories ...
2c171a43-082d-4e6f-b27e-e4680a7fb739
trentmkelly/LessWrong-43k
LessWrong
Humans: Not Carved from Marble Michael Vassar has been known to say that humans are not 'corrupted' by heuristics and biases and other elements of modern psychology. Humans just are psychology. Robert Kurzban puts this rather eloquently in his new book: > Michelangelo is famously quoted as saying, "I saw the angel i...
06d561e4-cb52-4d92-8736-561ffac0f1e8
trentmkelly/LessWrong-43k
LessWrong
Zvi's Law of No Evidence Update: Zvi wrote his own post. ---------------------------------------- > Law of No Evidence: Any claim that there is “no evidence” of something is evidence of bullshit. > > ―Covid 5/13: Moving On by Zvi > But in probability theory, absence of evidence is always evidence of absence > > ―...
547bf427-5019-41eb-ba12-6efdc71713ac
trentmkelly/LessWrong-43k
LessWrong
5. Moral Value for Sentient Animals? Alas, Not Yet Part 5 of AI, Alignment, and Ethics. This will probably make more sense if you start with Part 1. TL;DR In Parts 1 through 3 I discussed principles for ethical system design, and the consequences for AIs and uploads, and in Part 4, I discussed a principled way for us...
f64b66be-e420-4b3d-b9aa-938b4ba84902
trentmkelly/LessWrong-43k
LessWrong
Lifeism in the midst of death tl;dr:  My grandpa died, and I gave a eulogy with a mildly anti-deathist message, in a Catholic funeral service that was mostly pretty disagreeable. I'm a little uncomfortable writing this post, because it's very personal, and I'm not exactly a regular with friends here.  But I need to g...
f7d2b6bb-cefc-4994-a2ff-b7015ce8a666
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
The Credibility of Apocalyptic Claims: A Critique of Techno-Futurism within Existential Risk TLDR: Predictions about apocalyptic AI parallel both historical and contemporary Christain Apocalyptic claims(which I take to be untrustworthy). This is distinct from apocalyptic climate change which is non-analogous to such r...
c472fd5c-7ca6-4200-b2fb-47b9ebf9d8e7
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Breaking down the training/deployment dichotomy TL;DR: Training and deployment of ML models differ along several axes, and you can have situations that are like training in some ways but like deployment in others. I think this will become more common in the future, so it's worth distinguishing which properties of trai...
a5ae6794-ba15-49a1-9e43-c5cb626b5532
trentmkelly/LessWrong-43k
LessWrong
Does Evolutionary Theory Imply Genetic Tribalism? Belief in genetic tribalism comes from the idea that we have natural genetic interests in those who share our genes. This seems superficially plausible. We have a natural interest in our children, don’t we? That interest is because we share genes, right? If so, we shou...
95f87986-bee0-48cd-a2a3-bb8dfe078db9
trentmkelly/LessWrong-43k
LessWrong
Meetup : Montreal LessWrong Meetup - More Biases and Biased Board Gaming Discussion article for the meetup : Montreal LessWrong Meetup - More Biases and Biased Board Gaming WHEN: 10 December 2012 06:30:00PM (-0500) WHERE: 655 President Kennedy, Montreal We discussed lots of biases last week, so this week we're goin...
159fc18a-ddc8-499a-8634-7f392a731719
trentmkelly/LessWrong-43k
LessWrong
Examples of Superintelligence Risk (by Jeff Kaufman)
1c23f67d-24ce-4369-a976-49ee221de48a
trentmkelly/LessWrong-43k
LessWrong
Meetup : Budapest Meetup Discussion article for the meetup : Budapest Meetup WHEN: 07 April 2012 05:30:00PM (+0100) WHERE: Szent Istvan ter 4-5, Budapest Meeting at California Coffee Company Basilica (coffee shop), Szent Istvan ter 4-5. link. Please come and bring friends. If you have questions, contact katyusha. ...
58291697-039d-4e63-971d-c89224fd605d
trentmkelly/LessWrong-43k
LessWrong
The Kindness Project AI Safety is a group project for us all. We need everyone to participate - the ESFPs to the INTJs! Capturing the essence and subtleties of core values needs input across a broad span of humanity. Assumption 1 - large language models will be the basis of AGI. Assumption 2 - One way to add the ab...
1f6f02ff-7ddf-4247-9e0f-8a4a094ca394
trentmkelly/LessWrong-43k
LessWrong
Do Scientists Already Know This Stuff? poke alleges: > "Being able to create relevant hypotheses is an important skill and one a scientist spends a great deal of his or her time developing. It may not be part of the traditional description of science but that doesn't mean it's not included in the actual social instit...
c3522a61-81ba-4bf6-8a6b-95cad09b0306
StampyAI/alignment-research-dataset/arbital
Arbital
Guide Meta tag for the start page of a [multi-page guide](https://arbital.com/p/327).
26faf9a3-32f2-4cdf-a64b-6052dc04535f
trentmkelly/LessWrong-43k
LessWrong
Are you a virtue ethicist at heart? Disclaimer: I am not a philosopher, so this post will likely seem amateurish to the subject matter experts.  LW is big on consequentialism, utilitarianism and other quantifiable ethics one can potentially program into a computer to make it provably friendly. However, I posit that m...
6adbd0d9-7cc7-48e8-aab5-cba596cd5d79
trentmkelly/LessWrong-43k
LessWrong
Potential alignment targets for a sovereign superintelligent AI I'd like to compile a list of potential alignment targets for a sovereign superintelligent AI. By an alignment target, I mean something like what goals/values/utility function we might want to instill in a sovereign superintelligent AI (assuming we've so...
168db0aa-9756-4fc2-839f-317a282ff5ae
trentmkelly/LessWrong-43k
LessWrong
Changing habits for open threads Related to: Unjustified ideas comment thread Epistemic effort: Started off as a comment to the above thread, but then I spent too much time rewording and decided for it to be it's own post. 30m total thought. ---------------------------------------------------------------------------...
abfbe8e3-0383-466c-894c-ff3af9786ecb
trentmkelly/LessWrong-43k
LessWrong
Gaia Network: a practical, incremental pathway to Open Agency Architecture Introduction The Open Agency Architecture (OAA) proposal by Davidad has been gaining traction as a conceptual architecture to allow humanity to benefit from the nearly boundless upsides of AI while rigorously limiting its risks. As quoted here,...
27cfbe1b-0e33-4d2f-88c0-8f8a5f152d9b
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Demons in Imperfect Search One day, a gradient descent algorithm ball was happily rolling down a high-dimensional surface hill. All it wanted was to roll as far down as possible. Unbeknownst to the ball, just off to the side was a steep drop-off - but there was a small bump between the ball and the drop-off. No matter...
f421787f-cf31-4a19-af02-83e73d443c86
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
272. A Very Non Technical Explanation of the Basics of Infra Bayesianism hello and welcome to session 272 in the AI safety.com reading group tonight we will be discussing a very non-technical explanation of the basics of infropriationism by David matoxy David Metallica studies mathematics at a university in Budapest a...
073ec37b-ae25-4214-9923-3290ddb0cc1e
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
A Löbian argument pattern for implicit reasoning in natural language: Löbian party invitations *Related to:* [*Löb's Lemma: an easier approach to Löb's Theorem*](https://www.lesswrong.com/posts/AkGvmJ6WE5sXjuwnC/loeb-s-lemma-an-easier-approach-to-loeb-s-theorem)*.* Natural language models are really taking off, and i...
cb4e27d7-6ad3-4ae8-9a65-dc72cfbc572e
StampyAI/alignment-research-dataset/agisf
AGI Safety Fund
Is Power-Seeking AI an Existential Risk? 1 Introduction --------------- Some worry that the development of advanced artificial intelligence will result in existential catastrophe---that is, the destruction of humanity’s longterm potential.111See e.g. [Yudkowsky (2008)](https://intelligence.org/files/AIPosNegFactor....
3f0f7617-17f7-40d2-9b15-ea27f36d3bf5
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Exploiting Newcomb's Game Show Suppose you regularly watch a popular TV show where contestants play a game just like [Newcomb's Problem](https://www.lesswrong.com/tag/newcomb-s-problem) - before each contestant walks out on stage, the show's host sets out two boxes: box A containing nothing, and box B containing $1,00...
e227518a-9a04-4eab-9cd3-5c0c96ff5861
StampyAI/alignment-research-dataset/arxiv
Arxiv
Conditional Imitation Learning for Multi-Agent Games. I Introduction --------------- Many important robotics applications naturally involve multiple agents, from assistive robotics to self-driving cars. New techniques in deep multi-agent reinforcement learning have led to breakthrough performance in many multi-agen...
f01c37e0-80b4-4f7e-a25b-c687c6ecbc05
StampyAI/alignment-research-dataset/lesswrong
LessWrong
How good is humanity at coordination? When EAs look at the history of nuclear weapons, their reactions tend to fall into two camps. The first camp (which I am inclined towards) is “Man, what a total mess. There were so many near misses, and people involved did such clearly terrible and risky things like setting up th...
6e6194c8-757c-4fed-84af-a99c2253b822
trentmkelly/LessWrong-43k
LessWrong
A Natural Explanation of Nash Equilibria (This article is cross-posted from my blog) Imagine a world in which agents may interact, but may not coordinate in an enforceable way. Agents may communicate, but there is no mechanism to hold another agent accountable to commitments reached while communicating. In this world...
c99b133f-6d40-4b84-9554-c23ca2aac409
trentmkelly/LessWrong-43k
LessWrong
[Interim research report] Taking features out of superposition with sparse autoencoders We're thankful for helpful comments from Trenton Bricken, Eric Winsor, Noa Nabeshima, and Sid Black.  This post is part of the work done at Conjecture.   TL;DR: Recent results from Anthropic suggest that neural networks represent...
5e6c49dc-1b19-49a8-aebd-02441fa2f7c2
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Sections 1 & 2: Introduction, Strategy and Governance .mjx-chtml {display: inline-block; line-height: 0; text-indent: 0; text-align: left; text-transform: none; font-style: normal; font-weight: normal; font-size: 100%; font-size-adjust: none; letter-spacing: normal; word-wrap: normal; word-spacing: normal; white-space...
1553e214-feee-4f3f-90c3-41f32fea7344
trentmkelly/LessWrong-43k
LessWrong
Chapter 69: Self Actualization, Pt 4 It was out of the corner of her eye that Hermione Granger saw it, a reflection on the polished metal of a statue at the junction of two corridors, a flash of gold, a flash of red, something like an image of fire; just for a moment she saw it, and then it was gone. She paused, puzz...
d9b44404-3b91-4476-a488-678e3435c622
trentmkelly/LessWrong-43k
LessWrong
Against Belief-Labels EDIT: Lumifer says: "...when you are talking to someone, presumably you want to communicate some meaning and saying "I'm a rationalist" communicates very little." This is true. Belief-labels are useful socially; as long as you don't come to internalize them, you should be fine. I would dissuade...
942bb89a-f40c-4940-a102-e3a8e00a754b
trentmkelly/LessWrong-43k
LessWrong
Is there such a thing as an impossible protein? This is something I’ve been thinking about since my synthesizability article. Let’s assume, given the base twenty amino acids that are naturally present in the human body, we have every possible permutation of them for up to 100 amino acids, stored in a box with pH 7.4 ...
6d8727ae-76d3-4f64-a0d9-06b67c6136ab
trentmkelly/LessWrong-43k
LessWrong
Reframing AI Safety as a Neverending Institutional Challenge Crossposed  from https://stephencasper.com/reframing-ai-safety-as-a-neverending-institutional-challenge/  Stephen Casper “They are wrong who think that politics is like an ocean voyage or a military campaign, something to be done with some particular end...
365e2602-bf5a-4dc7-a160-220df4133508
trentmkelly/LessWrong-43k
LessWrong
Deception as the optimal: mesa-optimizers and inner alignment This is a brief distillation of Risks from Learned Optimization in Advanced Machine Learning Systems (Hubinger et al. 2019) with a focus on deceptive alignment. Watching The OTHER AI Alignment Problem: Mesa-Optimizers and Inner Alignment helped me better ...
fb9c409f-f17c-46dc-9e8f-d654a82702bc
trentmkelly/LessWrong-43k
LessWrong
Meetup : Effective Altruism Netherlands, cancelled Discussion article for the meetup : Effective Altruism Netherlands, cancelled WHEN: 15 February 2015 02:00:00PM (+0100) WHERE: Rijnkade 5, Utrecht We have informal meetups nearly biweekly, please see meetup.com: http://www.meetup.com/LWEANL/ Discussion article fo...
7016ed70-6d22-4afb-9d64-6ca4b1f49ef9
trentmkelly/LessWrong-43k
LessWrong
The "best predictor is malicious optimiser" problem Suppose you are a friendly AI A and have a mysterious black box B. B outputs a sequence of bits. You want to predict the next bits that B will output. Fortunately, you have a magic Turing machine oracle O. You can give O any computable function f( Turing machine, doe...
c709901e-b546-46b7-8a1c-94fc776b27c9
trentmkelly/LessWrong-43k
LessWrong
Papers, Please #1: Various Papers on Employment, Wages and Productivity For a while, I’ve been keeping a bookmark folder called ‘Papers, Please’ of all the papers I’d like to check out in the future. For those I do get to look at, I’ve compiled my observations, with the intent of making this another kind of roundup. I...
d16fb17d-caf2-4d52-88e6-3ab1d9ddfc73
trentmkelly/LessWrong-43k
LessWrong
More detailed proposal for measuring alignment of current models Here’s a more detailed proposal for a project someone could do on measuring alignment in current models. The aim is to build a good metric of how often a model is behaving in a misaligned way. If the project goes well, this could become a metric tha...
2d69a984-f8c8-450a-b85d-1b314f0ddc7b
trentmkelly/LessWrong-43k
LessWrong
Would a post about optimizing physical attractiveness be fitting for this forum? I've been allocating a fair amount of my time to this pursuit over the past year and have learned a lot about the subject and would love to share my experiences.
a99838b9-9fc6-40fa-925e-65aea555ac44
trentmkelly/LessWrong-43k
LessWrong
Rational and irrational infinite integers Epistemic status: a fine line between genius and madness * Let's start with the observation that it is much easier to use digits before the decimal point than the digits after it. A six years old child could calculate e.g. 11 + 2, but calculating reliably 0.11 + 0.2 will tak...
85e4e41d-3a06-4562-9165-d1bd6b6ba994
trentmkelly/LessWrong-43k
LessWrong
A Personal Record, and a Call For Advice: Introduction: This is my first post here on LessWrong, so I'll start off by introducing myself. My name is Caleb, and I've been reading here on LessWrong for quite some time now. I suspect nobody will be surprised when I say it was HPMOR that brought me here, as it has so ma...
ad1819f1-f9ae-4641-94dd-fb75ad53c572
trentmkelly/LessWrong-43k
LessWrong
The Friendly Drunk Fool Alignment Strategy Let's set some context!   A Philosophic Frame (half ironic, or maybe level two sarcasm) Systematicity isn't all that. I mean to say: formal systems only go so far.  Physicists gave up on having a single physics, so they just use two of them that contradict in general b...
e9525737-57ce-42d7-bb43-c044ad490cf4
StampyAI/alignment-research-dataset/arxiv
Arxiv
Scrutinizing and De-Biasing Intuitive Physics with Neural Stethoscopes 1 Introduction --------------- In the deep learning community, intuitive physics describes the concept of training neural networks to solve physics-related tasks in a data-driven as opposed to rule-based manner. However, what kind of function th...
ce56629b-3c54-416b-ab41-659df3fe1868
trentmkelly/LessWrong-43k
LessWrong
A model of the final phase: the current frontier AIs as de facto CEOs of their own companies The idea that the AI takes over its own company is obviously not a new one. For example, it's part of what happens in Joshua Clymer's "How AI Takeover Might Happen in 2 Years".  What's new (for me) is to take this very serio...
84b8847b-cf12-4367-b22e-36be4a05d8c5
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
The Alignment Newsletter #3: 04/23/18 **Highlights** -------------- **[Incomplete Contracting and AI Alignment](https://arxiv.org/abs/1804.04268)** *(Dylan Hadfield-Menell et al)*: This paper explores an analogy between AI alignment and incomplete contracting. In human society, we often encounter principal-agent prob...
d98d0653-41f3-4c4f-a0d5-051e80cf718d
StampyAI/alignment-research-dataset/arbital
Arbital
Reflexive relation A binary [relation](https://arbital.com/p/3nt) over some set is **reflexive** when every element of that set is related to itself. (In symbols, a relation $R$ over a set $X$ is reflexive if $\forall a \in X, aRa$.) For example, the relation $\leq$ defined over the real numbers is reflexive, because ...
8cb4cee1-2a58-428d-8598-c362d3fc2a4b
trentmkelly/LessWrong-43k
LessWrong
Good interview with Kahneman [link] http://uctv.ucsd.edu/search-details.aspx?showID=12295 Direct audio link: http://podcast.uctv.tv/mp3/12295.mp3 One thing that I found fascinating is the way he describes his collaboration with Tversky; he basically said that he had a "common mind" with his collaborator that was si...
b4856d62-be10-4b7f-ace0-0adb0e2ea975
trentmkelly/LessWrong-43k
LessWrong
AI risk, new executive summary AI risk Bullet points * By all indications, an Artificial Intelligence could someday exceed human intelligence. * Such an AI would likely become extremely intelligent, and thus extremely powerful. * Most AI motivations and goals become dangerous when the AI becomes powerful. * It i...
5f931430-91d6-45c7-abd0-adb48e6c6e0e
StampyAI/alignment-research-dataset/arbital
Arbital
Bayes' rule: Definition Bayes' rule is the mathematics of [probability theory](https://arbital.com/p/1rf) governing how to update your beliefs in the light of new evidence. [https://arbital.com/p/toc:](https://arbital.com/p/toc:) ## [Notation](https://arbital.com/p/1y9) In much of what follows, we'll use the follow...
83109f93-820a-4ab8-a0e2-b20c2895c5fb
StampyAI/alignment-research-dataset/arxiv
Arxiv
Avoiding Wireheading with Value Reinforcement Learning 1 Introduction --------------- As [Bostrom, 2014b](#bib.bib5) convincingly argues, it is important that we find a way to specify robust goals for superintelligent agents. At present, the most promising framework for controlling generally intelligent agents is ...
28d858b6-6ffc-4323-8115-ac27a02bf76f
trentmkelly/LessWrong-43k
LessWrong
Complete Wire Heading as Suicide and other things I came to the idea after a previous lesswrong topic discussing nihilism, and its several comments on depression and suicide. My argument is that wire heading in its extreme or complete/full form can be easily modeled as suicide, or less strongly as volitional intellige...
4aa0cd44-011b-4f82-aaa2-5a62fd50f901
trentmkelly/LessWrong-43k
LessWrong
The Fallacy of Dressing Like a Winner   Imagine you are a sprinter, and your one goal in life is to win the 100m sprint in the Olympics. Naturally, you watch the 100m sprint winners of the past in the hope that you can learn something from them, and it doesn't take you long to spot a pattern.   Every one of them ca...
895678e7-cbcc-4c52-9d20-0ca9aa7e14de
trentmkelly/LessWrong-43k
LessWrong
Stanislav Petrov has died (2017-05-19)
59952be5-bbb8-4aaf-a489-43cbd1ec6eee
trentmkelly/LessWrong-43k
LessWrong
Boston Solstice 2023 Retrospective Saturday evening we held another year's secular solstice celebration ( 2022, 2020, 2019, 2018), and I was again in my music director-ish role. Skyler was the main organizer this time around, with lots of support from Taymon. Scheduling was a bit tricky, as it always is. The weekends...
1f9fa423-a448-4d9e-82b2-c6ea7bb314bb
StampyAI/alignment-research-dataset/arxiv
Arxiv
Deep Adaptive Multi-Intention Inverse Reinforcement Learning 1 Introduction --------------- The task of learning from demonstrations (LfD) lies in the heart of many artificial intelligence applications [[28](#bib.bib28), [37](#bib.bib37)]. By observing the expert’s behavior, an agent learns a mapping between world ...
32d23587-2c22-49e0-86ad-3df39b4cd6c0
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
11 Major AI Developments: RT-2 to '100X GPT-4' there were 11 major developments this week in Ai and each one probably does deserve a full video but just for you guys I'm gonna try to cover it all here rt2 to scaling gpt4 100x stable Beluga 2 2 Senate testimony but let's start with rt2 which as far as I'm conce...
42358f6f-d5cb-45e1-a2c6-8160161cf648
trentmkelly/LessWrong-43k
LessWrong
Link: The Cook and the Chef: Musk's Secret Sauce - Wait But Why This is the fourth of Tim Urban's series on Elon Musk, and this time it's about some reasoning processes that are made explicit, which LW readers should find very familiar. It's a potentially useful explicit model of how to make decisions for yourself. h...
d6be4aab-2410-46b7-a816-31ca70c15073
trentmkelly/LessWrong-43k
LessWrong
Kingfisher Tour February 2025 Last week Kingfisher went on tour with Alex Deis-Lauby calling. Similar plan to last year: February break week, rented minivan, same caller, many of the same dances and hosts. This time our first dance was Baltimore, and while it's possible to drive from Boston to Baltimore in one day an...
bce1156f-af26-4b4b-8b28-58511eb6eedc
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Basic Mathematics of Predictive Coding This is an overview of the classic 1[999 paper by Rajesh Rao and David Ballard](https://www.nature.com/articles/nn0199_79) that introduced predictive coding. I'm going to focus on [explaining the mathematical setup instead of just staying at a conceptual level](https://www.lesswr...
c03a63ee-c3b8-49eb-bd2e-31f3468dff90
trentmkelly/LessWrong-43k
LessWrong
Meetup : Urbana-Champaign: Planning and Re-planning Discussion article for the meetup : Urbana-Champaign: Planning and Re-planning WHEN: 20 April 2014 12:00:00PM (-0500) WHERE: 412 W. Elm St, Urbana, IL When things get complicated enough, you have to plan them in advance or they fail. You need blueprints and logist...
33872f9f-a698-431b-a4db-8e7c4597059a
trentmkelly/LessWrong-43k
LessWrong
Course selection based on instructor I've been trying to compile advice on how to select good instructors in college, assuming that the major and course level are given. Of course, many people don't have a lot of flexibility with instructor choice, so this advice has limited applicability. The reason I'm posting to Le...
dfe21836-c3bc-4f7d-a4fb-431873e2a7e6
trentmkelly/LessWrong-43k
LessWrong
An Audio Introduction to Nick Bostrom Team Radio Bostrom has just released An Audio Introduction to Nick Bostrom. It offers a deep dive into Nick Bostrom's ideas, including: existential risk, wise philanthropy, the ethics of AI, transhumanism, and the case for speeding up some investments in technology while slowing ...
e5f0e0c9-dec4-4a1f-9b1d-8c3618731f2f
StampyAI/alignment-research-dataset/blogs
Blogs
Saving The Client-Side Web: just WASM and the DOM Saving The Client-Side Web: just WASM and the DOM ------------------------------------------------- the stack of technologies involved in the client side of the web is a huge mess, and i don't like most of it. it requires browsers to implement basically an entire ope...
77bd4a72-6a7b-4da5-b97f-33e4b2bcaf71
trentmkelly/LessWrong-43k
LessWrong
HPMoR Podcast Over 9000 = New Trailer The Harry Potter and the Methods of Rationality audiobook/podcast exceeded nine thousand total downloads last night! Over 800 per episode on average. In celebration I put together another trailer. :) This one is taken from Chapter 7, the most popular chapter so far. I like to thin...
a1c71ed2-7626-4144-b1cb-1b4774fa45dc
trentmkelly/LessWrong-43k
LessWrong
Fiction: Tasmanian Devil ch. 1 Tasmanian Devil is a story of college love, mystery, magic, logic and ethics. It parodises the human nature to the core. It can get very dark and is based loosely on rationality. Read Chapter 1 here. (PG-13; warnings: very dark themes, a few vulgar words)  Teaser: > Baby tooth was a m...
a71460a2-69b3-4060-908a-aa0900fc7797
trentmkelly/LessWrong-43k
LessWrong
EU’s AI ambitions at risk as US pushes to water down international treaty (linkpost) From EU’s AI ambitions at risk as US pushes to water down international treaty – EURACTIV.com (June 6, 2023) As summarized by Claude: * The Council of Europe is drafting an international treaty on AI called the Convention on Artifi...
d0ba1626-8f3c-4ecf-9722-4c045db11d2a
trentmkelly/LessWrong-43k
LessWrong
Meetup : Monday Madison Meetup Discussion article for the meetup : Monday Madison Meetup WHEN: 17 October 2011 06:00:00PM (-0500) WHERE: 1831 Monroe St, Madison, WI WHERE: Barriques on Monroe, 1831 Monroe St, Madison, WI WHEN: 6pm Monday, October 17 I plan to bring the stuff we'd need to play Zendo, which is actual...
3a258264-a968-4d78-b273-89ff96f73207
trentmkelly/LessWrong-43k
LessWrong
Imitative Generalisation (AKA 'Learning the Prior') Tl;dr We want to be able to supervise models with superhuman knowledge of the world and how to manipulate it. For this we need an overseer to be able to learn or access all the knowledge our models have, in order to be able to understand the consequences of suggestio...
9e9629d3-1eb7-4d65-89bf-eefb1c940224
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post91 Overview: By training neural networks with selective modularity, gradient routing enables new approaches to core problems in AI safety. This agenda identifies related research directions that might enable safer development of transformative AI. Introduction Soon, the world may see rapid increases in AI capa...
01216623-7d39-4a7a-92bb-5561bd5fbc24
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Supporting global coordination in AI development: Why and how to contribute to international AI standards *(Link post for the FHI GovAI technical report [Standards for AI Governance: International Standards to Enable Global Coordination in AI Research & Development](https://www.fhi.ox.ac.uk/wp-content/uploads/Standard...
979fe78f-fbc9-4c69-b97d-606e79160ceb
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Absence of Evidence Is Evidence of Absence Title: [SEQ RERUN] Absence of Evidence Is Evidence of Absence Tags: sequence_reruns Today's post, Absence of Evidence Is Evidence of Absence was originally published on 12 August 2007. A summary (taken from the LW wiki): > Absence of proof is not proof of absence...
ebfcd349-6449-491b-9133-4704018032e5
trentmkelly/LessWrong-43k
LessWrong
[Book Review] The Chrysanthemum and the Sword Japan was the only non-Western[1] country to build an industrial empire before the establishment of the liberal world order. The Chrysanthemum and the Sword: Patterns of Japanese Culture by Ruth Benedict is thus a real-world case study of what culture could have beenen in ...
ba845e0a-dc33-434d-bfa1-0ce41a11727e
trentmkelly/LessWrong-43k
LessWrong
Be Nice to Non-Rationalists Note: I have no intention of criticizing the person involved. I admire that (s)he made the "right" decision in the end (in my opinion), and I mention it only as an example we could all learn from. I did request permission to use his/her anecdote here. I'll also use the pronoun "he" when rea...
ce04cfba-d807-4adb-991f-935ea1b6b17d
trentmkelly/LessWrong-43k
LessWrong
Agreeing With Stalin in Ways That Exhibit Generally Rationalist Principles > It was not the sight of Mitchum that made him sit still in horror. It was the realization that there was no one he could call to expose this thing and stop it—no superior anywhere on the line, from Colorado to Omaha to New York. They were in ...
559723e2-5c9f-437a-9cb5-851a68ca8076
StampyAI/alignment-research-dataset/lesswrong
LessWrong
A Corrigibility Metaphore - Big Gambles I present here a helpful analogy to understand the[*corrigibility problem*](https://www.lesswrong.com/tag/corrigibility) and the challenge raised by MIRI in their [proposal](https://intelligence.org/files/Corrigibility.pdf). This analogy simplifies greatly some challenges of cor...
8f29f91d-6af6-47c1-b06f-d1f433f7d1a3
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] The Majority Is Always Wrong Today's post, The Majority Is Always Wrong was originally published on April 3, 2007. A summary (from the LW wiki): > Anything worse than the majority opinion should get selected out, so the majority opinion is rarely strictly superior to existing alternatives. Discuss the p...
d77f8997-c2d3-4465-8329-0400218a6028
trentmkelly/LessWrong-43k
LessWrong
Meetup : Baltimore Discussion article for the meetup : Baltimore WHEN: 07 May 2017 08:00:31PM (-0400) WHERE: Performing Arts and Humanities Building, 1000 Hilltop Cir, Baltimore, MD 21250, United States Casual discussion group! Suggested Reading: [http://lesswrong.com/lw/ouc/project_hufflepuff_planting_the_flag/]. ...
37a91aac-0b56-4689-b453-2a5be1335157
trentmkelly/LessWrong-43k
LessWrong
Security Mindset - Fire Alarms and Trigger Signatures Series Overview and Goals This is the second in a series of articles about applying traditional security mindset to the problems of alignment and AI research in general. As much as possible, we should try to mine the lessons from the history of security and apply t...
4e825fc7-020e-4a09-90a1-5571b515380d
trentmkelly/LessWrong-43k
LessWrong
What Programming Language Characteristics Would Allow Provably Safe AI? It seems clear that many high-level programming languages are candidates for use in the first AGI. They have enough power to write that code. It seems clear, however, that the power that those languages have is incompatible with formal safety. SPA...
d0b8c86c-06e0-46ee-99df-eee7e1c3bfee
trentmkelly/LessWrong-43k
LessWrong
SAE features for refusal and sycophancy steering vectors TL;DR * Steering vectors provide evidence that linear directions in LLMs are interpretable. Since SAEs decompose linear directions, they should be able to interpret steering vectors. * We apply the gradient pursuit algorithm suggested by Smith et al to decompo...
a249b19c-dc00-4f6c-918a-f4ea1e7c30bd
trentmkelly/LessWrong-43k
LessWrong
CEA Disambiguation In 2011 effective altruists in Oxford had two main organizations: Giving What We Can and 80,000 Hours. They wanted to incorporate, and created the Centre for Effective Altruism as an umbrella organization to host them: [1] Over time they started running other projects: conferences, supporting ...
dfae0895-4f08-4de1-a46c-d5f65721ae65
trentmkelly/LessWrong-43k
LessWrong
Should one be sad when an opportunity is lost? There are many ways to tackle this question, but I mean this in a homo economicus, not biased perspective. If we were great optimizers of some things (experiences, states of the world, utility in the emotional sense), should we be sad upon hearing we lost an opportunity? ...
e3bed2ae-5121-4fb7-8155-6064203e0d25
trentmkelly/LessWrong-43k
LessWrong
San Francisco Meetup 4/4 7:00 PM Monday April 4th at 7:00 PM Green Papaya 825 Mission Street (4th and Mission)San Francisco, CA 94103 Announcing the creation of a third Bay Area LW meetup, San Francisco! The first meeting of the San Francisco group will be this coming Monday, April 4th, from 7:00-9:00 at Green Papay...
8595457b-b4a1-4fd7-8142-30be03087d01
trentmkelly/LessWrong-43k
LessWrong
Roadmap for a collaborative prototype of an Open Agency Architecture Deger Turan, Tantum Collins, davidad and Charbel-Raphael Segerie contributed to the writing of this post.  Andrew Critch, Ben Goldhaber, Ozzie Gooen, and Evan Miyazono reviewed this post.  tl;dr: Open Agency Architecture (OAA) is a bold proposal fo...
36b38fe6-8bf1-427c-a10c-fb41030b719f
trentmkelly/LessWrong-43k
LessWrong
They don't make 'em like they used to Recently someone on Twitter posted a picture like this, commenting that this kind of stove still works after a hundred years, but “thanks to ‘progress’ and ‘improvement’ you have to replace your new one after 5 years.” Columbian Stove Works No. 28 “Modern Mistress,” Ulster Americ...
a6fae695-49a2-474b-9015-bb6744c48e13
trentmkelly/LessWrong-43k
LessWrong
The prior of a hypothesis does not depend on its complexity Many thanks to Unknowns for inventing the scenario that led to this post, and to Wei Dai for helpful discussion. Imagine you subscribe to the universal prior. Roughly, this means you assign credence 2^-k to each program of length k whose output matches your ...
eb653a69-8194-4121-867b-26766d238515
trentmkelly/LessWrong-43k
LessWrong
Game Theory and Partner Abuse I wrote a reflection on Game Theory and partner abuse. Curious about this community's thoughts. *** I confess: I’ve been getting sucked into the Johnny Depp/Amber Heard trial. Last week devolved into a blur of articles, Twitter threads, and subreddits, culminating in the testimony of th...
3c0fcc8d-6c51-41b5-ada4-f197590012d7
trentmkelly/LessWrong-43k
LessWrong
Shard theory alignment has important, often-overlooked free parameters. A delayed hot take. This is pretty similar to previous comments from Rohin. Shard theory alignment requires magic - not in the sense of magic spells, but in the technical sense of steps we need to remind ourselves we don't know how to do. Locatin...
69f62056-0227-4345-9469-3941ad0abf97
trentmkelly/LessWrong-43k
LessWrong
February 2021 gwern.net newsletter None
baaf2662-0c89-42f7-8e8f-9dd6848b9501
StampyAI/alignment-research-dataset/arxiv
Arxiv
Uncertainty Quantification for Competency Assessment of Autonomous Agents I Introduction --------------- While real world deployment of autonomous robotic agents will greatly benefit humans, the challenge lies in establishing appropriate levels of trust that lead to proper usage of and reliance on the autonomy [[10...