id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
13aa8654-08ef-45ea-aead-ea1f44677ba9
StampyAI/alignment-research-dataset/arxiv
Arxiv
Layer Normalization 1 Introduction --------------- Deep neural networks trained with some version of Stochastic Gradient Descent have been shown to substantially outperform previous approaches on various supervised learning tasks in computer vision [Krizhevsky et al., [2012](#bib.bib1)] and speech processing [Hinto...
eac9738f-882f-41a6-bfc6-565264bfc992
trentmkelly/LessWrong-43k
LessWrong
What mathematics to learn There is, of course, Kahn Academy for fundamentals. We have already had a discussion on How to learn math. What resources exist detailing which mathematics to learn in what order? What resources exist that explain the utility of different mathematical subfields for the purpose of directing s...
c255ce77-a1b6-4dfe-817c-e3c6d2241c14
trentmkelly/LessWrong-43k
LessWrong
AI Safety Discussion Days AI Safety Discussion Days I’m going to organise an online AI Safety discussion day, roughly once per month. This is a space to discuss anything you want related to preventing catastrophic risk from AI, including governance, strategy, all technical AI Safety research, and also trying to under...
49d00d9a-15ff-4583-b732-4f6dd443065a
trentmkelly/LessWrong-43k
LessWrong
Designing Artificial Wisdom: Decision Forecasting AI & Futarchy Introduction In this post I will describe one possible design for Artificial Wisdom (AW.) This post can easily be read as a stand-alone piece, however it is also part of a series on artificial wisdom. In essence: Artificial Wisdom refers to artificial in...
854e9dfa-20e9-4f87-84b0-fbdb6b20df50
trentmkelly/LessWrong-43k
LessWrong
Rationality Quotes June 2012 Here's the new thread for posting quotes, with the usual rules: * Please post all quotes separately, so that they can be voted up/down separately.  (If they are strongly related, reply to your own comments.  If strongly ordered, then go ahead and post them together.) * Do not quote your...
b33876c2-52b8-4970-a962-f17f70b655a4
trentmkelly/LessWrong-43k
LessWrong
Social Insight: When a Lie Is Not a Lie; When a Truth Is Not a Truth - Pt. 2 '"If people have a right to be stupid, the market will respond by supplying all the stupidity that can be sold."'  People misinterpret this as indicating that I take a policy stance in favor of regulation.  It indicates no such thing.  It is ...
c7078cdc-3b79-492a-8ebd-e0bb9c229388
trentmkelly/LessWrong-43k
LessWrong
Whiteboard Program Traceing: Debug a Program Before you have the Code TL;DR: A Whiteboard Program Trace is where you write down a data structure on a whiteboard and step-by-step walk through how an algorithm A would manipulate this data, i.e. you execute A for one specific input in a visual way. Usually, I do this bef...
2036170a-fa8b-400e-ba38-327d26e48a9d
trentmkelly/LessWrong-43k
LessWrong
The Next Right Token After many repeats of Frozen II's "The Next Right Thing" in preparation for secular solstice, figuring out the chords and practicing it with Julia, it struck me that doing the next right thing is an awful lot like next token prediction. So here's a filk from the perspective of an LLM at the end of...
2b45f08f-e7d4-4c4e-b526-00b71fd0b911
trentmkelly/LessWrong-43k
LessWrong
Found Paper: "FDT in an evolutionary environment" I'm not sure what to think of this paper, it's quite long and I haven't finished checking it for sanity. nevertheless, I noticed it hadn't made its way here, and there are mighty few papers that cite the FDT paper, so I figured I'd drop it off rather than leave it sitt...
241f2ea1-8529-494d-8ed4-a79b2d978021
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
Reinforcement Learning 7: Planning and Models today I wanted to talk models but in the context of learning maybe I should have put it in the title but just this quickly to recap what we've been doing last lecture we learned policies directly from experience using policy gradients methods and such before that we talked...
22f66f51-489a-409e-a97d-4b6ebf0e75b3
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
PreDCA: vanessa kosoy's alignment protocol (*this post has been written for the second [Refine](https://www.alignmentforum.org/posts/5uiQkyKdejX3aEHLM/how-to-diversify-conceptual-alignment-the-model-behind) blog post day. thanks to* [*vanessa kosoy*](https://www.lesswrong.com/users/vanessa-kosoy)*,* [*adam shimi*](htt...
9bd15aad-d876-4d93-afab-3e08b2220005
trentmkelly/LessWrong-43k
LessWrong
Comprehensive COVID-19 Disinfection Protocol for Packages and Envelopes An easy way to limit your exposure to COVID-19 is to quit going out grocery shopping and buy groceries online instead. But there's a problem: This review found that coronaviruses can persist on inanimate surfaces for up to 9 days at room temperatu...
afbe6509-63a6-481f-80a9-ca23d0418f4b
trentmkelly/LessWrong-43k
LessWrong
Microsoft and OpenAI, stop telling chatbots to roleplay as AI AI demos should aim to enhance public understanding of the technology, and in many ways ChatGPT and Bing are doing that, but in one important way they aren't: by appearing to talk about themselves. This creates understandable confusion and in some cases fea...
b1f19ad3-535f-40fd-82e4-079287278648
trentmkelly/LessWrong-43k
LessWrong
The president didn't die: failures at extending AI behaviour A putative new idea for AI control; index here. In a previous post, I considered the issue of an AI that behaved "nicely" given some set of circumstances, and whether we could extend that behaviour to the general situation, without knowing what "nice" reall...
42c7fc0e-1728-4985-bb21-3b471dddb294
trentmkelly/LessWrong-43k
LessWrong
Paradigm shifts in forecasting This post has been written in relation with work I'm doing for the Machine Intelligence Research Institute (MIRI), but hasn't been formally vetted by MIRI. I'm posting this to LessWrong because of its potential interest to a segment of the LessWrong readership. As always, all thoughts ar...
df95ce5d-7c98-4304-af7e-a28b823075b8
trentmkelly/LessWrong-43k
LessWrong
Predicting AGI by the Turing Test Abstract This essay explains the Direct Approach proposed by [@barnettScalingTransformativeAutoregressive2023].[1] I encourage you to play with the Direct Approach Interactive Model to explore an interactive simulation using the approach. > The Direct Approach framework bounds the co...
8c5368f5-b5f8-4ccb-8d8c-ad102aded883
trentmkelly/LessWrong-43k
LessWrong
Progress links and tweets, 2023-06-21: Stewart Brand wants your comments Opportunities * Dwarkesh Patel is hiring a COO for his podcast, The Lunar Society * Join a new campaign for nuclear regulatory reform (via @BuildNuclearNow) Announcements * Stewart Brand is serializing his book on Maintenance and inviting yo...
de5e7c98-58f0-47bb-99d7-d88eb0b0b6c7
trentmkelly/LessWrong-43k
LessWrong
Chapter 95: Roles, Pt 6 The third meeting (10:31am, April 17th 1992) Spring had begun, the late-morning air still crisp with the leavings of winter. Daffodils had bloomed amid the sprouting grass of the forest, the gentle yellow petals with their golden hearts dangling limply from their dead, grayed stems, wounded or...
c47183f1-cdae-478b-9a1f-d232acf931fb
trentmkelly/LessWrong-43k
LessWrong
Trajectories to 2036 TL;DR  This post follows on from an earlier, longer attempt to think through Holden Karnofsky's ‘AI Could Defeat All of Us Combined’ (AIDC).  It spends less time on the specifics of precisely how a mass-copied human-level AI (HLAI) would kill, enslave, or disempower humans (i.e. who would copy the...
c0946a6e-9ce1-41bd-8944-ab1ee6503c16
trentmkelly/LessWrong-43k
LessWrong
Risk-Free Bonds Aren't I've always been annoyed by the term "risk-free bonds rate", meaning the return on US Treasury bills.  Just because US bonds have not defaulted within their trading experience, people assume this is impossible?  A list of major governments in 1900 would probably put the Ottoman Empire or Austria...
6d9aa765-64f3-461e-9547-896710d27e69
StampyAI/alignment-research-dataset/lesswrong
LessWrong
GPT-3, belief, and consistency I've seen a few people debating what GPT-3 understands, and how this compares to human understanding. I think there's an easier and more fruitful question to consider: what does it believe? It seems like it doesn't believe anything, or alternately, it believes everything. It's a categor...
62fc158d-a969-4dad-81f8-73e9f21e78bf
trentmkelly/LessWrong-43k
LessWrong
Linkpost: Memorandum on Advancing the United States’ Leadership in Artificial Intelligence > Memorandum on Advancing the United States’ Leadership in Artificial Intelligence; Harnessing Artificial Intelligence to Fulfill National Security Objectives; and Fostering the Safety, Security, and Trustworthiness of Artificia...
f1a3a2c7-03b5-4da9-9117-d5e8d633f5b9
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Non-superintelligent paperclip maximizers are normal The [paperclip maximizer](https://en.wikipedia.org/wiki/Instrumental_convergence) is a thought experiment about a hypothetical superintelligent AGI that is obsessed with maximizing paperclips. It can be modeled as a utility-theoretic agent whose utility function is ...
588b1ee3-8508-47cd-87f6-6728afeb8b8f
trentmkelly/LessWrong-43k
LessWrong
RA Bounty: Looking for feedback on screenplay about AI Risk Jamie Wahls and Arthur Frost are writing a screenplay for Rational Animations. It's a sci-fi episodic comedy illustrating various ways in which AI goes off the rails in more or less catastrophic ways.  We're looking for feedback and offering bounties. This i...
92a10e93-5e9c-4cbe-a25d-6e9457f07654
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post3767 A putative new idea for AI control; index here . One of the ways in which humans aren't agents is that we can be manipulated into having any (or almost any) set of values - through drugs, brain surgery, extreme propaganda, and other methods. If an AI is tasked with "satisfy human preferences", and if the ...
127e57e7-5b21-4db6-8e1a-af1f52b5ecdd
trentmkelly/LessWrong-43k
LessWrong
Weekly LW Meetups: Stockholm, Dublin, Penn State, Cambridge MA, Ottawa, and Marin CA There are upcoming irregularly scheduled Less Wrong meetups in: * Stockholm meetup: 09 October 2011 02:00PM * Dublin, IE, Meetup: 15 October 2011 02:00PM * Penn State University: NEW Meetup starting!: 22 October 2011 02:00PM The ...
652b3414-680c-4fe3-bdbf-5ef4f5b81f5d
StampyAI/alignment-research-dataset/blogs
Blogs
Neuron firing rates in humans Our best guess is that an average neuron in the human brain transmits a spike about 0.1-2 times per second. Support ------- ### Bias from neurons with sparse activity When researchers measure neural activity, they can fail to see neurons which rarely fire during the experiment (thos...
89db16aa-e1ae-4ab9-9ba1-f407e47c4e94
trentmkelly/LessWrong-43k
LessWrong
Google DeepMind's RT-2 Abstract > We study how vision-language models trained on Internet-scale data can be incorporated directly into end-to-end robotic control to boost generalization and enable emergent semantic reasoning. Our goal is to enable a single end-to-end trained model to both learn to map robot observatio...
a80b81e4-01fd-4b43-a532-1163ba31931b
trentmkelly/LessWrong-43k
LessWrong
Theories of Regret (The following is armchair psychological speculation based on anecdotal evidence. If anyone can respond with relevant science, that'd be awesome, but otherwise responses in kind are quite welcome.) I'm puzzled regarding the motivational effects of negative emotion, particularly shame, guilt, and re...
22557ccb-e4fb-4e4e-a040-04d5c64a906d
trentmkelly/LessWrong-43k
LessWrong
Supposing you inherited an AI project... Supposing you have been recruited to be the main developer on an AI project. The previous developer died in a car crash and left behind an unfinished AI. It consists of: A. A thoroughly documented scripting language specification that appears to be capable of representing any ...
0c6434e6-ed85-475b-882a-0015df51fc9f
StampyAI/alignment-research-dataset/special_docs
Other
Quantilizers: A safer alternative to maximizers for limited optimization Quantilizers: A Safer Alternative to Maximizers for Limited Optimization Jessica Taylor Machine Intelligence Research Institute jessica@intelligence.org Abstract In the eld of AI, expected utility maximizers are commonly used as a model for idea...
16d47fd3-ed9b-4e3d-88f4-fd91d7e9d34e
trentmkelly/LessWrong-43k
LessWrong
The AI Explosion Might Never Happen [This is a crosspost from https://amistrongeryet.substack.com/p/recursive-self-improvement-foom, lightly edited for the LessWrong audience. This is my first LessWrong post; feedback greatly appreciated!] LessWrong readers will be familiar with the concept of recursive self-improvem...
e37bc549-82ea-4317-b797-eeb8b5898ff6
trentmkelly/LessWrong-43k
LessWrong
Reestablishing Reliable Sources: A System for Tagging URLs Before diving in, I will preface saying I have never written a post before, and am in general a lurker. However this idea has been itching on the back of my mind for months and I would love feedback on what I may be missing or if this is something to seriously...
0ee95a38-e4fb-4523-a2f5-a05454cb7638
trentmkelly/LessWrong-43k
LessWrong
The Rational Investor, Part I First off, note that I am not in possession of any of the licenses that entitle me to call myself a financial advisor. However, the approach to investing I will present in this article is endorsed by many economists, Warren Buffet, and Vanguard. I personally follow it, and you can too. W...
a81af44b-2dd2-4755-b015-c1da275d3a63
trentmkelly/LessWrong-43k
LessWrong
If I Was An Eccentric Trillionaire cross-posted from niplav.site > What I might do if I magically got hold of a very large amount of money, and couldn't spend it maximally altruistically. I sometimes like to engage in idle speculation. One of those speculations is: "If someone came up to me and told me that they wou...
a1a5c1c3-0bed-454e-b446-d42ea39ec18d
trentmkelly/LessWrong-43k
LessWrong
Venture Granters, The VCs of public goods, incentivizing good dreams Epistemic Status: Early concept The Gist of it: To support sane risktaking, an institution should generally grant career capital in proportion to a quantitative record of the harms and benefits of a decisionmaking entity's past funding decisions. An...
0a37a57e-1bf6-4a69-8842-dee599da1d1e
trentmkelly/LessWrong-43k
LessWrong
[LINK] Why taking ideas seriously is probably a bad thing to do Yvain's blog: Epistemic learned helplessness. A friend in business recently complained about his hiring pool, saying that he couldn't find people with the basic skill of believing arguments. That is, if you have a valid argument for something, then you s...
94bcc0fa-84e8-472e-ba45-3a509b7cc6d6
StampyAI/alignment-research-dataset/arxiv
Arxiv
Friendly Artificial Intelligence: the Physics Challenge I The Friendly AI vision ------------------------- As Irving J. Good pointed out in 1965 [Good65](#bib.bib2) , an AI that is better than humans at all intellectual tasks could repeatedly and rapidly improve its own software and hardware, resulting in an “intel...
e9861e2f-d9cf-4a95-b003-fdd811b9a538
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Technical AGI safety research outside AI I think there are many questions whose answers would be useful for technical AGI safety research, but which will probably require expertise outside AI to answer. In this post I list 30 of them, divided into four categories. Feel free to get in touch if you’d like to discuss the...
26801b10-ccd7-43cc-8a34-446e806aeef1
trentmkelly/LessWrong-43k
LessWrong
Decoherence is Pointless Previously in series:  On Being Decoherent Yesterday's post argued that continuity of decoherence is no bar to accepting it as an explanation for our experienced universe, insofar as it is a physicist's responsibility to explain it.  This is a good thing, because the equations say decoherence...
4d687204-a137-4d4f-8965-875e0bbd9423
StampyAI/alignment-research-dataset/arbital
Arbital
Axiom of Choice: Introduction "The Axiom of Choice is necessary to select a set from an infinite number of pairs of socks, but not an infinite number of pairs of shoes." — *Bertrand Russell, Introduction to mathematical philosophy* "Tarski told me the following story. He tried to publish his theorem \[equivalence be...
4b74bac6-43e6-494a-94fa-0cc49f623cb6
trentmkelly/LessWrong-43k
LessWrong
Half-Baked Products and Idea Kernels When I ask someone at work for a project proposal, I never want the person to go silent on me and put in 100 hours of solitary work, and then finally show me something and ask for my feedback. I always want to see a half-baked product. You can half-bake something in an hour or two...
55c34f71-eaaf-4c21-a42e-b314b4397472
trentmkelly/LessWrong-43k
LessWrong
Two kinds of Agency Original post: http://bearlamp.com.au/two-kinds-of-agency/ ---------------------------------------- Let's talk about agency.  This week I read Transform Your Self, as reviewed by Kaj Sotala in How I found and fixed the root problems behind my depression and anxiety after 20 years. Transform your...
fbc92ba4-9969-4e0a-8c3f-4703916b0fb7
trentmkelly/LessWrong-43k
LessWrong
Agentic Interpretability: A Strategy Against Gradual Disempowerment Authors: Been Kim, John Hewitt, Neel Nanda, Noah Fiedel, Oyvind Tafjord Full paper on arXiv   We propose a research direction called agentic interpretability. The idea of agentic interpretability stems from the observation that AI systems are becom...
67912113-6ce2-4142-b61f-c5063ac7f638
trentmkelly/LessWrong-43k
LessWrong
Six Months of ROSE Six months ago, we launched the Guild Beta. Our goal was -- and is -- to produce people who are impressive in everyday life. We are interested in real-life effectiveness, not armchair philosophy. This attitude is why, when people join the Guild and embark on our Rank Advancement program, we have the...
d2968875-3100-4c5a-9e87-813dda05532f
trentmkelly/LessWrong-43k
LessWrong
Open thread, Mar. 27 - Apr. 02, 2017 Notes for future OT posters: 1. Please add the 'open_thread' tag. 2. Check if there is an active Open Thread before posting a new one. (Immediately before; refresh the list-of-threads page before posting.) 3. Open Threads should start on Monday, and end on Sunday. 4. Unflag th...
8e9f8aaa-bb71-4d89-a654-3511b4107460
trentmkelly/LessWrong-43k
LessWrong
Are unpaid UN internships a good idea? Disclaimer: I am outside of the world of international organisations. I am a scientific researcher at university. I am writing this post to open a discussion. Introduction UN is an international organisation with the following main goals: * maintain international peace and se...
046212fa-0d05-4178-b9ee-90223634a5de
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"In Keep Your Identity Small, Paul Graham argues against associating yourself with labels (i.e. “libertarian,” “feminist,” “gamer,” “American”) because labels constrain what you’ll let yourself believe. It’s a wonderful essay that’s led me to make concrete changes in my life. That said, it’s only about 90% correct. I h...
49022a36-da8d-4cc0-8c5e-77966cf8ac9c
trentmkelly/LessWrong-43k
LessWrong
What is "wireheading"? This is an article in the featured articles series from AISafety.info. AISafety.info writes AI safety intro content. We'd appreciate any feedback.  The most up-to-date version of this article is on our website, along with 300+ other articles on AI existential safety. Wireheading happens when a...
2f497844-f513-400d-99a6-02df1901d5e7
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
[Intro to brain-like-AGI safety] 14. Controlled AGI *Part of the* [*“Intro to brain-like-AGI safety” post series*](https://www.alignmentforum.org/s/HzcM2dkCq7fwXBej8)*.* 14.1 Post summary / Table of contents ===================================== [Post #12](https://www.alignmentforum.org/posts/Sd4QvG4ZyjynZuHGt/intro...
25b15e06-09d0-4a87-a9ca-04f02d8df547
trentmkelly/LessWrong-43k
LessWrong
What are examples of simpler universes that have been described in order to explain a concept from our more complex universe? Sometimes there's a concept that can be difficult to understand when entangle with everything else that needs to be understood about our physics. If you isolate that concept in a simpler unive...
19386a15-5f0a-4f2d-a4c0-9399c20160f2
trentmkelly/LessWrong-43k
LessWrong
Book review: Trick or treatment (2008) TL;DR: Read if you enjoy witnessing applied rationality.   In their book from 2008, Simon Singh and Edzard Ernst went through the evidence for and against a large range of alternative, a.k.a. complementary, health interventions. And wowza, was that an interesting journey!   ...
890f4d11-688a-4fa0-a096-f02ec2c6e908
trentmkelly/LessWrong-43k
LessWrong
Some thoughts on CBDC The CBDC is beyond a payment innovation facing customers. What it really changes is the lending relationships between big financial institutions (wholesale) and customers (retail). Let us say you hold some retail CBDC, it will be a central bank's liability, instead of a commercial bank's liabilit...
2852326c-a3e7-4f62-9618-3cad97c636ff
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Reward function learning: the learning process In the [previous post](https://www.lesswrong.com/posts/55hJDq5y7Dv3S4h49/reward-function-learning-the-value-function), I introduced the formalism for reward function learning, and presented the expected value function for a learning agent: V(μ,ρ,π,hn)=∑hm∈Hm∑R∈Rm∑i=1Pπ,μ...
09f024aa-7dda-42c6-98b8-015c224cd9b9
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Inner alignment in the brain **Abstract:** We can think of the brain crudely as (1) a neocortex which runs an amazingly capable quasi-general-purpose learning-and-planning algorithm, and (2) subcortical structures (midbrain, etc.), one of whose functions is to calculate rewards that get sent to up the neocortex to dir...
670d2ca2-8c51-4750-b1e3-d9c44138a6d2
trentmkelly/LessWrong-43k
LessWrong
Article upvoting Except in rare cases (like Wei Dai's Fair Division of Black-Hole Negentropy) I'm still using article upvotes to partially determine whether to promote articles to the front page - some informal mixture of "number of upvotes" + "editor's judgment".  I mention this because while comment voting is still ...
e275eff9-9c89-4c6b-a4e6-787323a9f333
trentmkelly/LessWrong-43k
LessWrong
Problems with forensics, including bias affecting DNA testimony Forensic Science and the Innocence Project > What’s more, some types of DNA technology have shortcomings, as Boise State University geneticist Greg Hampikian cautioned Philadelphia meeting attendees. Sample collection methods haven’t changed since DNA’s ...
1d2acd05-cdfc-4a07-b9af-6e3a8e88c2b8
trentmkelly/LessWrong-43k
LessWrong
Alignment from equivariance II - language equivariance as a way of figuring out what an AI "means" I recently had the privilege of having my idea criticized at the London Institute for Safe AI, including by Philip Kreer and Nicky Case. Previously the idea was vague; being with them forced me to make the idea specific....
828d04c3-7178-45ba-80bb-ac50a8b1d859
trentmkelly/LessWrong-43k
LessWrong
D&D.Sci Alchemy: Archmage Anachronos and the Supply Chain Issues Evaluation & Ruleset This is a follow-up to last week's D&D.Sci scenario: if you intend to play that, and haven't done so yet, you should do so now before spoiling yourself. There is a web interactive here you can use to test your answer, and generation...
75687bf1-8327-4228-a712-3aac7efb6a2b
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
If I were a well-intentioned AI... IV: Mesa-optimising .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-spac...
27d79f90-1ea6-4ac6-a494-dd8b61b5d682
trentmkelly/LessWrong-43k
LessWrong
200 COP in MI: Interpreting Reinforcement Learning This is the ninth post in a sequence called 200 Concrete Open Problems in Mechanistic Interpretability. Start here, then read in any order. If you want to learn the basics before you think about open problems, check out my post on getting started. Look up jargon in my...
d456a880-2c91-46a3-9feb-74efcf831a5c
trentmkelly/LessWrong-43k
LessWrong
Skoll World Forum: Catastrophic Risk and Threats to the Global Commons More: Skoll Global Threats Fund | To Safeguard Humanity from Global Threats > The panel surfaced a number of issues that contribute to our inability to date to make serious strides on global challenges, including income inequality, failure of gove...
127d81e2-897f-4f07-9b3f-b2b97b3f1b29
trentmkelly/LessWrong-43k
LessWrong
The Context is Conflict Cross-posted from Putanumonit.com. ---------------------------------------- John Nerst wrote an excellent analysis of the recent clash between Sam Harris and Ezra Klein on the topic of race, IQ, and all the topics that get dragged in when you talk about race and IQ. The essay weaves together ...
204b9c4a-7dc2-4ede-aa41-313f5ff7e951
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Once upon a time, I met someone who proclaimed himself to be purely selfish, and told me that I should be purely selfish as well. I was feeling mischievous(*) that day, so I said, "I've observed that with most religious people, at least the ones I meet, it doesn't matter much what their religion says, because whatever...
8bd915d2-5cdb-4f11-a71b-48647df90e07
trentmkelly/LessWrong-43k
LessWrong
Decision Theories, Part 3.75: Hang On, I Think This Works After All Followup to: Decision Theories, Part 3.5: Halt, Melt and Catch Fire, Decision Theories: A Semi-Formal Analysis, Part III The thing about dead mathematical proofs is that it's practically always worth looting the corpse; sometimes you even find there'...
04828ea5-a130-491f-a871-ae0c0a6cd6d2
trentmkelly/LessWrong-43k
LessWrong
LW/OB Quotes - Fall 2009 This is a monthly thread for posting any interesting rationality-related quotes you've seen on LW/OB. * Please post all quotes separately (so that they can be voted up/down separately) unless they are strongly related/ordered. * Do not quote yourself. * Do not post quotes that are NOT com...
5c75dec6-d8c0-4f29-b74c-fc5745ede521
trentmkelly/LessWrong-43k
LessWrong
How to Lurk Less (and benefit others while benefiting yourself) Posts like this have been written before, but I think it's worth making the point periodically. Lurker ratios have likely increased over time. Comments and discussion are an important feedback mechanism for content creators. So if you see stuff you like,...
4301a30b-3280-4d77-9fc3-ad8963c23f0c
trentmkelly/LessWrong-43k
LessWrong
SIAI vs. FHI achievements, 2008-2010 After reading the FHI achievement report for 2008-2010, I thought it might be useful to compare their achievements to those of SIAI during the same time period. Since SIAI does not have an equivalent report, I've mostly pulled the data of their achievements from the SIAI blog. My ...
1ed99d99-cd63-47a9-8fed-2ec3b808e22a
trentmkelly/LessWrong-43k
LessWrong
How much of a concern are open-source LLMs in the short, medium and long terms? I don't mean only doomist concerns, I mean in general. What are specific things in which these open-source models might be an extra-problem?   Thank you
e6e1d74e-6af9-4205-a017-8a7cbc1f0954
trentmkelly/LessWrong-43k
LessWrong
How Do Induction Heads Actually Work in Transformers With Finite Capacity? Thanks to Marius Hobbhahn for helpful discussions and feedback on drafts. I think the usual explanation of how induction heads work in actual Transformers, as described in Anthropic’s Mathematical Framework, is incomplete. Concretely, it seems...
4cae6107-78f2-4466-831b-a71d3481c53b
trentmkelly/LessWrong-43k
LessWrong
Where Would Good Forecasts Most Help AI Governance Efforts? Thanks to Josh Rosenberg for comments and discussion. Introduction One of LessWrong’s historical troves is its pre-ChatGPT AGI forecasts. Not just for the specific predictions people offered, but for observing which sorts of generative processes produced wh...
83aebae7-49c3-42f2-b223-109afd3c139c
trentmkelly/LessWrong-43k
LessWrong
Transformative VR Is Likely Coming Soon Meta (formerly known as Oculus and as Facebook) just had their yearly conference. I think right now is a good time to take a brief break from worrying about AI timelines, and take a few minutes to notice the shortness of VR timelines. VR isn't going to determine whether humanity...
876ff7be-e62d-4c62-bd73-4abc68f453b7
trentmkelly/LessWrong-43k
LessWrong
N=1 Self Analysis The N=1 scientific field of self-study doesn't seem to have much established basis. It mostly seems to exist through self-experimentation of drugs by through non-conventional channels (i.e. mostly non-academic). It seems incredibly useful to me, but very awkward. Our brains are powerful, and arguabl...
56fe68c8-7990-4825-a913-90488d14e929
trentmkelly/LessWrong-43k
LessWrong
Replaceable Axioms give more credence than irreplaceable axioms Epistemic status: hand waving but confidant. I was thinking about reflective reasoning and "strange loops through the meta level" and it lead me to this intuition that even if you have to make some assumptions to support your beliefs, if your beliefs can...
fd17f6e1-51cd-421c-a4dd-572ef3ceaaa2
trentmkelly/LessWrong-43k
LessWrong
Are there any groupchats for people working on Representation reading/control, activation steering type experiments? Looking for any discord/slack/other that have people working on projects related to representation reading, control, activation steering with vectors and adapters, ...Would appreciate any pointers if su...
c801b2e8-ecf3-43e0-8397-96179446c89d
trentmkelly/LessWrong-43k
LessWrong
Tune Your Cognitive Strategies None
7b152f81-ae6f-4f2c-9885-08f0b704d970
trentmkelly/LessWrong-43k
LessWrong
A brief theory of why we think things are good or bad Moral reasoning depends on believing that some things are good and others are bad. Some of these things seem very believable - I am quite firmly convinced that intense pain is bad. Though they seem very compelling, it’s hard to point to strong evidence that proves ...
e85553fa-58c9-43eb-8db4-9380059ff6cd
trentmkelly/LessWrong-43k
LessWrong
Why Do We Believe Summary: The fundamental question of rationality is, why do you believe what you believe? In this exercise, we ask that question repeatedly in an attempt to figure out our foundations. Tags: Repeatable, small, experimental Purpose: The purpose is twofold. Firstly, we want to grow comfortable and fa...
f3ca23eb-5da8-4692-94db-6b94ab2b31a1
trentmkelly/LessWrong-43k
LessWrong
[Link] Ignorance, a skilled practice https://carcinisation.com/2020/01/27/ignorance-a-skilled-practice/ Excerpt: The Global Knowledge Game To illustrate that global knowledge is a game, consider a story about Alexander Luria, who studied illiterate Russian peasants and their semi-literate children. Consider especia...
5dd1814e-d4be-4fb9-8af8-f31a58b52c28
trentmkelly/LessWrong-43k
LessWrong
PSA: Very important policy change at Cryonics Institute In the past, the Cryonics Institute has had a policy that said that they would not accept anyone who is not a member. This has changed. The policy now is that someone who has full legal authority over your body can sign you up after you die. It costs $36,250 to b...
3ca6da5f-593f-4706-a618-9a8f5952522a
trentmkelly/LessWrong-43k
LessWrong
Public Static: What is Abstraction? Author’s Note: Most of the posts in this sequence are essentially a log of work-in-progress. This post is intended as a more presentable (“public”) and higher-confidence (“static”) write-up of some formalizations of abstraction. Much of the material has appeared in other posts; the ...
153f6c38-ea57-4e5d-b778-52c9ff28f637
trentmkelly/LessWrong-43k
LessWrong
Papers to start getting into NLP-focused alignment research I have recently started working towards a research-focused MSc in AI under the supervision of Dr. Pilehvar and Dr. Soleymani at Sharif University of Technology. (CV, LinkedIn) Since we essentially get no funding here in Iran, I have ample freedom in choosing...
a261ab79-662c-4fab-a5b6-9ac4af03d18d
trentmkelly/LessWrong-43k
LessWrong
If MWI is correct, should we expect to experience Quantum Torment? If the many worlds of the Many Worlds Interpretation of quantum mechanics are real, there's at least a good chance that Quantum Immortality is real as well: All conscious beings should expect to experience the next moment in at least one Everett branch...
616f91cc-55f2-46e1-b11c-e1aeba81575f
trentmkelly/LessWrong-43k
LessWrong
Automatic programming, an example Say, that we have the following observational data:   Planet Aphelion 000 km Perihelion 000 km Orbit time days Mercury 69,816 46,001 88 Venus 108,942 107,476 225 Earth 152,098 147,098 365 Mars 249,209 206,669 687 Jupiter 816,520 740,573 4,332 Saturn 1,513,325 1,353,572 10,760 Uranus...
6729d001-b9f5-4834-8d41-a01b62e5acd7
trentmkelly/LessWrong-43k
LessWrong
The Artificial Intentional Stance Another post in the same incremental vein. Still hoarding the wild speculation for later. I The idea of the "intentional stance" comes from Dan Dennett, who wanted to explain why it makes sense that we should think humans have such things as "beliefs" or "desires." The intentional s...
fe490488-145f-4ddc-b45a-f53d0488ac7c
trentmkelly/LessWrong-43k
LessWrong
Why do firms choose to be inefficient? Firms choose inefficient production technologies for seemingly bizarre reasons. I examine the evidence, and show that this genuinely is worse for everyone, not merely the result of unobserved constraints. I think this can be explained by a model where information is cheap to atta...
3765541c-b744-4b56-accf-2bc4b49be7d5
trentmkelly/LessWrong-43k
LessWrong
Seeking Collaborators I've been accepted as a mentor for the next AI Safety Camp. You can apply to work with me on the tiling problem. The goal will be to develop reflectively consistent UDT-inspired decision theories, and try to prove tiling theorems for them. The deadline for applicants is November 17. The program...
ba0a554c-8e9f-4944-9a39-176304e52217
trentmkelly/LessWrong-43k
LessWrong
Debate helps supervise human experts [Paper] There didn't seem to be a link post to this recent paper on AI debate yet, so I figured I would make one:  > As AI systems are used to answer more difficult questions and potentially help create new knowledge, judging the truthfulness of their outputs becomes more difficul...
761f3aa3-5df1-4e19-bb46-65708b21bf20
trentmkelly/LessWrong-43k
LessWrong
The Solution is Inaction I'm trying to get my current mental model of the Covid crisis in writing, to see both where it leads and where it is incorrect. This is an economic model. Section 1 is defining the problem and vocabulary, section 2 is a description of my mental model, and section 3 is comparing my model to rea...
cd0f477c-bf7a-40bc-aa37-36ec9f8705f6
trentmkelly/LessWrong-43k
LessWrong
Hacking the CEV for Fun and Profit It’s the year 2045, and Dr. Evil and the Singularity Institute have been in a long and grueling race to be the first to achieve machine intelligence, thereby controlling the course of the Singularity and the fate of the universe. Unfortunately for Dr. Evil, SIAI is ahead in the game....
ae0c3474-ff0e-4f6a-98e6-a055d2e0fd56
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Shutting down AI Safety Support This is a quick post to officially announce that AI Safety Support is on an indefinite pause. We have been on pause for a while now, but I wanted to make it official. I might have more to say on this later but expect that if I wait till I write a longer explanation, it will not happen. ...
55eeec94-1dac-40c9-af1d-953c3f50e302
trentmkelly/LessWrong-43k
LessWrong
What does davidad want from «boundaries»? Chipmonk As the Conceptual Boundaries Workshop (website) is coming up, and now that we're also planning Mathematical Boundaries Workshop in April, I want to get more clarity on what exactly it is that you want out of «boundaries»/membranes. So I just want to check: Is your go...
2b08a0d8-952d-445a-ba50-8de738b20233
trentmkelly/LessWrong-43k
LessWrong
Was the historical Jesus talking about proto-evolution? (You might be surprised) Out of all the research rabbit holes I've ever gone down, this one is by far my favorite, as to most people at first glance it's so unthinkable an idea. Years ago, I would have been right there with you in disbelief, but the reasons why ...
18a009e8-86f1-42e4-947b-302f6d3cad1b
trentmkelly/LessWrong-43k
LessWrong
Time Magazine has an article about the Singularity... ...and it is surprisingly positive. Here is the Cover Here is the article online The money quote, IMO, given how mainstream a publication this is: "The difficult thing to keep sight of when you're talking about the Singularity is that even though it sounds like...
666f04b5-7afd-4894-9533-996c04b7f010
trentmkelly/LessWrong-43k
LessWrong
Training Trace Priors and Speed Priors Thanks to Evan Hubinger for suggesting this idea. Training Trace Priors are priors over boolean circuits which examine the outputs of gates on samples from the training distribution, typically for purposes of steering models away from having components that were never tested dur...
d3e472d9-6b79-440c-b954-a9fabd022741
trentmkelly/LessWrong-43k
LessWrong
Falsehoods you might believe about people who are at a rationalist meetup I go to a lot of rationalist meetups. I quite enjoy them, and it’s often because of the people who go to the meetups. There’s a number of assumptions you might have about people who go to rationalist meetups, and many of them are mostly true. Ho...
6d9b507d-5ee9-48a1-bf1b-bae044933bad
trentmkelly/LessWrong-43k
LessWrong
Meetup : Moscow Meetup: Codename Felix Discussion article for the meetup : Moscow Meetup: Codename Felix WHEN: 14 September 2014 02:00:00PM (+0400) WHERE: Russia, Moscow, ulitsa L'va Tolstogo 16 Here's our plan: * A talk about Gollwitzer, Implementation intentions and Rubicon model. * Calibration excercise. * D...
7978988b-1175-490f-961c-83801fadd987
StampyAI/alignment-research-dataset/arbital
Arbital
Logical game In the context of [AI Alignment Theory](https://arbital.com/p/2v), a 'logical' game is one that we are, for purposes of thought experiment, treating as having only the mathematical structure of the game as usually understood. In real-world chess, you can potentially bribe the opposing player, drug them, ...
d2c058be-31d9-4dc3-81fa-3634f6e2c8dc
trentmkelly/LessWrong-43k
LessWrong
ChatGPT refuses to accept a challenge where it would get shot between the eyes [game theory] Yesterday it was Jaws and game theory, today it’s Sir Gawain and the Green Knight (henceforth SGGK). But really, the sequence runs in the opposite direction. As you may know, SGGK is a medieval romance that starts and ends in ...
97981ae5-ab65-4e7f-adf0-651869ae4682
trentmkelly/LessWrong-43k
LessWrong
Reflecting on the 2022 Guild of the Rose Workshops Overview The Guild of the Rose is a community of learners dedicated to personal growth and development, often with a rationalist bent. In 2022, the Guild offered a variety of workshops to help its members hone their skills and become more knowledgeable in a variety of...