id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
96f00118-13db-41e2-8198-0fc747250ecf
StampyAI/alignment-research-dataset/aisafety.info
AI Safety Info
What are some good podcasts about AI alignment? - The [AI section](https://80000hours.org/topic/world-problems/most-pressing-problems/artificial-intelligence/?content-type=podcast) of the 80,000 Hours Podcast - The [AI X-risk Research Podcast](https://axrp.net/), hosted by Daniel Filan - [The Inside View,](http://th...
5692ffaa-c66f-430b-ba75-e41d47a248c1
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Value learning for moral essentialists Many people - people we want to persuade - are essentialists about morality and ethics. They give weight to the idea that just like how knowing facts about my shoes is possible because my shoes exist and interact with me, facts about what is right or good are possible because the...
4a59285f-190d-4503-925c-c34d6d98db9a
trentmkelly/LessWrong-43k
LessWrong
How is low-latency phasic dopamine so fast? (Quick poorly-researched blog post, probably only of interest to neuroscientists.) There’s a paper “What is reinforced by phasic dopamine signals?” (Redgrave, Gurney, Reynolds 2008). I was interested in it as part of my project to continue checking whether I should really b...
d56e22d7-835c-4053-b4b1-17a63cd0ea88
trentmkelly/LessWrong-43k
LessWrong
Homeostatic Bruce Epistemic Status: Speculative. CW: trauma Humans are Adaptation Executors, Not Fitness Maximizers. — some guy 0 Summary Bruce is a weird guy, and I’m not the first to hypothesize that he serves an adaptive purpose. What I hope to add to the discussion is to distinguish between optimistic and pess...
9393fdaf-8ff7-4713-98bf-9082f91157b7
trentmkelly/LessWrong-43k
LessWrong
Collapsing “Collapsing the Belief/Knowledge Distinction” TL;DR: After re-examining my earlier stance on collapsing the belief/knowledge distinction, I've come to realize through Wittgenstein's "meaning as use" that these terms hold distinct meanings based on their contextual applications in language. Embracing this un...
5e975f84-1681-4e44-b576-c95f19b49707
trentmkelly/LessWrong-43k
LessWrong
Quantum AI Box With quantum computing, an AI box needs not guard against hardware exploits. Epistemic status: armchair computer science Here's a simple machine model no more powerful than a quantum computer. It's a computer with extra instructions. It can spend x*t time to, within the machine, split time into x² ti...
2215b10a-dce4-4e90-a7d1-64401cad17cf
trentmkelly/LessWrong-43k
LessWrong
Recommendations on communities that discuss AI applications in society I find that most communities I am part of, where AI discussions occur, fall into two categories: 1. The community is too broad, and the discussion is fragmented. ACX would fall into this category. 2. The AI discussion focuses on AGI/ASI and rel...
57fee2bc-57fb-4df1-ae3e-03a456c6c112
trentmkelly/LessWrong-43k
LessWrong
Duckbill Masks Better? Lai et al. just published a paper looking at how well various masks worked for keeping others from getting sick ("source control"). Their university press office summarized this as: > Study finds all masks effective, but 'duckbill' N95 masks far outperform others, suggests they should be the ...
7ca8d8e2-866b-4088-bcb9-dc360c397409
trentmkelly/LessWrong-43k
LessWrong
$1000 USD prize - Circular Dependency of Counterfactuals Congrats to the winner TailCalled with their post Some thoughts on "The Nature of Counterfactuals". See the winner announcement post. I've previously argued that the concept of counterfactuals can only be understood from within the counterfactual perspective. ...
ec7763b9-5a7c-4832-bd57-94c5da9e7203
trentmkelly/LessWrong-43k
LessWrong
How To Actually Succeed Epistemic Status:  Casual musings on Eliezer's LessWrong post "Trying to Try" Slightly different though similar YouTube version: Also viewable on my website: https://jordanarel.com/2022/09/12/how-to-actually-succeed/ And The EA Forum https://forum.effectivealtruism.org/posts/nbvcbB4iC8J2...
34f33079-6407-458d-8c52-ec8071ae219b
trentmkelly/LessWrong-43k
LessWrong
A diagram for a simple two-player game (Copied from my blog) I always have a hard time making sense of preference matrices in two-player games. Here are some diagrams I drew to make it easier. This is a two-player game: North wants to end up on the northernmost point, and East on the eastmost. North goes first, an...
72771072-aa74-46e6-8e12-2a3282fa56c9
trentmkelly/LessWrong-43k
LessWrong
[updated] how does gpt2′s training corpus capture internet discussion?  not well [Updated to correct my earlier claim that this doesn't affect GPT-3. Apparently it does?] I’m out sick today, but had enough energy to do some GPT-related fiddling around. This time, I was curious what “internet discussions” tended to l...
0bbea69d-80c6-4067-97b1-059ab8eb1a0e
trentmkelly/LessWrong-43k
LessWrong
2025 Alignment Predictions I’m curious how alignment researchers would answer these two questions: 1. What alignment progress do you expect to see in 2025? 2. What results in 2025 would you need to see for you to believe that we are on track to successfully align AGI?
113130b9-f0cd-4d6b-8bbf-ec7e791c0ca0
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Followup to: Talking Snakes: A Cautionary Tale Related to: Explain, Worship, Ignore Skepticism is like sex and pizza: when it's good, it's very very good, and when it's bad, it's still pretty good.It really is hard to dislike skeptics. Whether or not their rational justifications are perfect, they are doing society a ...
8c2a70c5-1042-43a4-8ff9-fdf87b2346f0
trentmkelly/LessWrong-43k
LessWrong
Communities you might join thread Follow up to the Archipelago model of community standards. If you are interested in different social norms or goals than the current community has, post a comment here. The idea would be to try something and see how it went whilst still keeping in touch with LW/EA sphere. I thought ...
b43402e7-33c3-4c40-9653-249c53ba0240
trentmkelly/LessWrong-43k
LessWrong
Meetup : Less Wrong NH Meetup Discussion article for the meetup : Less Wrong NH Meetup WHEN: 14 June 2016 07:00:00PM (-0400) WHERE: 269 Pearl St Manchester NH 03104 The twentieth NH meet-up is Tuesday, 6/14, in Manchester, NH at 7 pm at a private residence. Light refreshments will be provided. Have you read Ration...
29e29078-9a5b-4abc-9011-f99a06b97c03
trentmkelly/LessWrong-43k
LessWrong
What are the best published papers from outside the alignment community that are relevant to Agent Foundations? I am after papers that you have stumbled across that may be relevant to the core of the understanding the existential AI alignment problem despite not having explicit links to AI alignment.  Here are papers...
f4176c75-4101-4fad-a287-6a649194fe78
trentmkelly/LessWrong-43k
LessWrong
[link] Bayesian inference with probabilistic population codes Bayesian inference with probabilistic population codes by Wei Ji Ma et al 2006 > Recent psychophysical experiments indicate that humans perform near-optimal Bayesian inference in a wide variety of tasks, ranging from cue integration to decision making to m...
caba6ae8-c31e-453d-b23a-fb40acd18c79
StampyAI/alignment-research-dataset/arbital
Arbital
Emphemeral premises A purported [psychological](https://arbital.com/p/43h) fallacy wherein people confronted with an incongruent proposition X will say "Oh, not-X because Y", and perhaps a week later, say, "Oh, not-X because Z." [Ephemeral Premises](https://www.facebook.com/yudkowsky/posts/10154268184344228) is when ...
95846c61-8802-4b7a-bc16-ce693d66b464
trentmkelly/LessWrong-43k
LessWrong
June 2016 Media Thread This is the monthly thread for posting media of various types that you've found that you enjoy. Post what you're reading, listening to, watching, and your opinion of it. Post recommendations to blogs. Post whatever media you feel like discussing! To see previous recommendations, check out the ol...
192e716f-8b51-42e0-852c-f07d633d3c02
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
The Terminology of Artificial Sentience *Edited by Jacy Reese Anthis and Ali Ladak. Many thanks to Thomas Moynihan, Tobias Baumann, and Teo Ajantaival for reviewing and providing feedback. This article is also available and referenceable on the Open Science Framework:* [*https://doi.org/1/0.31234/osf.io/sujwf*](https:...
517bd54e-763a-4c65-9adc-f462b7a465a3
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Lotteries: A Waste of Hope Today's post, Lotteries: A Waste of Hope was originally published on April 13, 2007. A summary (from the LW wiki): > Some defend lottery-ticket buying as a rational purchase of fantasy — paying a dollar for a day's worth of pleasant anticipation. But then your valuable brain is ...
a45b66be-8750-43b2-bb39-113ed47033b1
trentmkelly/LessWrong-43k
LessWrong
Defense Against the Dark Prompts: Mitigating Best-of-N Jailbreaking with Prompt Evaluation This is a linkpost for a new research paper of ours, introducing a simple but powerful technique for preventing Best-of-N jailbreaking. Abstract > Recent work showed Best-of-N (BoN) jailbreaking using repeated use of random au...
259764a8-4228-4236-998c-ce2ed792782e
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Life and expanding steerable consequences *Financial status: This is independent research. I welcome* [*financial support*](https://www.alexflint.io/donate.html) *to make further posts like this possible.* *Epistemic status: I believe this is a helpful lens through which to view the significance of AI in a way that ...
ae58d9f6-8146-49dd-ba26-ac3e8c89ce7a
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
The economy as an analogy for advanced AI systems Eric Drexler’s Comprehensive AI Services (CAIS), particularly as set out in his 2019 report [*Reframing Superintelligence*](https://www.fhi.ox.ac.uk/wp-content/uploads/Reframing_Superintelligence_FHI-TR-2019-1.1-1.pdf), is a complex model with many different assumption...
9acec390-b00f-4720-8ad8-b6fcc95103b0
trentmkelly/LessWrong-43k
LessWrong
Nerdtrition: simple diets via spreadsheet abuse I found several simple (four to seven total foods) diets that, with carefully specified amounts of each food, appear, at face value, to provide full, proper nutrition. Here are a few, with amounts given per day. * fish/olive/etc * 450 g cooked tilapia * 450 g rip...
c5732d08-7f7f-4e3a-b050-3da803b15991
trentmkelly/LessWrong-43k
LessWrong
My Notes on the book, "Turn the ship around" by L David Marquet
9e56d365-e26b-4da8-b55a-04100e113a08
trentmkelly/LessWrong-43k
LessWrong
Subjective Realities So I have a friend who I sit next to in class, and we talk about philosophy. Well today, he brought up that when people leave your presense, and you can't observe them any longer, you no longer have proof they exist. Well I pointed out that it would violate the conservation of mass law, right? S...
0227a647-2dde-4df1-90c8-1df57276e841
trentmkelly/LessWrong-43k
LessWrong
How good is the case for retraining yourself to sleep on your back? I frequently hear the advice that it's better to sleep on the back and worthwhile to learn to sleep on your back. Are there any studies that backup that advice. Otherwise are there other good arguments? Personal experience is also welcome.
1263abff-da0c-43cd-9e08-59431382207d
StampyAI/alignment-research-dataset/arxiv
Arxiv
Making Responsible AI the Norm rather than the Exception Montreal AI Ethics Institute An international, non-profit research institute helping humanity define its place in a world increasingly driven and characterized by algorithms Website ​ : ​ https://montrealethics.ai Newsletter ​ : ​ https://a...
7429f9a1-ac33-409a-969e-e981be34a0fc
trentmkelly/LessWrong-43k
LessWrong
Prediction Market News/Ticker? Lately I’ve been finding myself wanting a way to track how prediction markets have been changing with time. Some markets are pretty stable, while others drift from week to week or even swing violently. These trends feel like relevant news that I want as part of my information diet. Are ...
9e534dd9-787e-4745-96f6-5e1369d6d240
trentmkelly/LessWrong-43k
LessWrong
Kurzweil's predictions' individual scores In my post on Kurzweil's 1999 predictions for 2019, I wrote: > The (anonymised) data can be found here, and I encourage people to download and assess it themselves. But it is, of course, not trivial for everyone to download a spreadsheet and wade through it, and you might wa...
c2680c07-243d-4b34-b31f-5d2f83b6962d
trentmkelly/LessWrong-43k
LessWrong
Write Good Enough Code, Quickly At the start of my Ph.D. 6 months ago, I was generally wedded to writing "good code". The kind of "good code" you learn in school and standard software engineering these days: object oriented, DRY, extensible, well-commented, and unit tested. I like writing "good code" - in undergrad I ...
f5b448cf-dde3-4055-9785-73361f213cde
trentmkelly/LessWrong-43k
LessWrong
Decision Theory Anti-realism This is crossposed from my blog. While I think the ideas here are solid I think the presentation still needs some work so I'd also appreciate comments on the presentation so I can turn this into a more polished essay, e.g., is the second section worth keeping and what should be expanded/cu...
e710de19-abf5-4a1e-9a15-6ae0d04952cf
trentmkelly/LessWrong-43k
LessWrong
Do-it-yourself-science Wiki While I was busy procrastinating, I produced http://38020.vs.webtropia.com/sciencewiki/index.php/Helium_balloon (the Wiki-Extension for the plot, not the data, that was lying around anyway). This could (if enough people are using it) become quite a useful collection of evidence of various s...
9affd8e7-a3d4-445f-989c-3c5f9aad762d
trentmkelly/LessWrong-43k
LessWrong
What types of compute/processing could we distinguish? I am a bit confused and thought I'd rather ask and discuss here before thinking about it for long. As usual I am trying to compartmentalize, structure, make distinctions. My confusion was triggered thinking about the evaluation function (heuristic to rate the cer...
2151a552-5fba-41f3-9699-780ad8efed9d
trentmkelly/LessWrong-43k
LessWrong
Forging A New AGI Social Contract This is the introductory piece for a series of essays written by AI economists, policy researchers, and political thinkers on the topic of a new AGI Social Contract.  Current contributors include: Anton Korinek, Deger Turan (CEO of Metaculus), Steve Omohundro, Iason Gabriel (DeepMind...
0defd4a7-eeb5-4f9d-9c49-b997f84a6474
trentmkelly/LessWrong-43k
LessWrong
I Want to Learn About Education I want to learn what's well-understood about education. I expect to launch myself into some endeavors in teaching the first few levels of epistemic and instrumental rationality - ie., critical thinking and problem solving. I'm a little suspicious, though, of the scattered educational te...
8782010d-a34b-498c-ab5e-ca1980a91c40
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
Evan Hubinger on Inner Alignment, Outer Alignment, and Proposals for Building Safe Advanced AI [Music] [Music] welcome to the ai alignment podcast i'm lucas perry today we have a conversation with evan hubenger about ideas and two works of his an overview of 11 proposals for building safe advanced ai and risks from le...
8058e67a-49ae-4711-8853-e3434a5355c2
trentmkelly/LessWrong-43k
LessWrong
Sleep, Diet, Exercise and GLP-1 Drugs As always, some people need practical advice, and we can’t agree on how any of this works and we are all different and our motivations are different, so figuring out the best things to do is difficult. Here are various hopefully useful notes. TABLE OF CONTENTS 1. Effectiveness...
80b26b0b-2915-4afc-ba53-a0c3bdafff17
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Google's ethics is alarming I published this video on my YouTube channel yesterday.  In short, [Google](https://wiki.tournesol.app/index.php/Google) is clearly [investing massively in AI performance](https://thenextweb.com/neural/2021/01/13/googles-new-trillion-parameter-ai-language-model-is-almost-6-times-bigger-tha...
f69fcbe8-f361-4acf-b53e-ce70ceb6cd98
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Why I Believe LLMs Do Not Have Human-like Emotions [*EDIT:* I don't know exactly why this received so many downvotes - whether it was my tone or arguments or both - but I accept that this was not, in the end, a strong post. I am a new writer here and I am still figuring out how to improve my argumentation. That said,...
9b3aa60c-f568-42a2-b8e6-590fc02db29c
trentmkelly/LessWrong-43k
LessWrong
Notes from Online Optimal Philanthropy Meetup: 12-10-09 Here are my notes from the Optimal Philanthropy online meeting. Things in square brackets are my later additions and corrections. Let me know if there are factual errors or if I've incorrectly captured the drift of what you were saying. Nick: * existential ris...
c920774c-050e-4531-b0df-e2ada7305fc8
trentmkelly/LessWrong-43k
LessWrong
[Link] Rethinking the way colleges teach critical thinking "Though critical thinking is universally regarded as a pillar of higher education ... results show that students are not developing their critical thinking skills to the extent we expect."   http://blogs.scientificamerican.com/guest-blog/2012/12/14/re-thinki...
daf2dcf8-c071-47cc-a7f2-097ff03b0027
trentmkelly/LessWrong-43k
LessWrong
Teenage Rationalists and Changing Your Mind I remember the moment when I became an atheist. I was reading Religion's Claim to Be Non-Disprovable, an uneasy feeling growing in my head, and then I reached the bottom of the article, stared at the screen for a couple of seconds, and got it. "There is no God," I whispere...
5f619e41-2764-48e6-8142-3ecad762c5c3
trentmkelly/LessWrong-43k
LessWrong
Some half-serious, half-formed thoughts on existing and so on So I’ve been banging my head against a wall (metaphorically, almost not) for about a week and a half (or years on and off) about the apparent meaninglessness of anything and the difficulty of finding anything to do that is mildly satisfying next to the absu...
edc32f45-67f7-4753-9d8e-49d5502700a9
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
What Does (and Doesn't) AI Mean for Effective Altruism? | Owen Cotton-Barratt | EAGxBerlin 2017 fine yeah I guess some of you may have noticed that within the eventual tourism community there's a reasonable amount of attention given towards artificial intelligence and I want to talk about a bit about that and what I t...
a14751c6-0752-4179-b507-6920e6d79d8c
trentmkelly/LessWrong-43k
LessWrong
The Unilateralist’s “Curse” Is Mostly Good People around here sometimes reference the “Unilateralist’s Curse”, especially when they want to keep someone else from doing something that might cause harm. Briefly, the idea is that “unilateral” actions taken without the consent of society at large are especially likely to...
91511e40-2563-490d-bd6b-bd4af46691df
StampyAI/alignment-research-dataset/lesswrong
LessWrong
How I'm thinking about GPT-N There has been a lot of hand-wringing about accelerating AI progress within the AI safety community since OpenAI's publication of their [GPT-3](https://arxiv.org/abs/2005.14165) and [Scaling Laws](https://arxiv.org/abs/2001.8361) papers. OpenAI's clear explication of scaling provides a jus...
298115a7-bb24-403e-a8b7-cb85e1f7ef7f
trentmkelly/LessWrong-43k
LessWrong
The Critical COVID-19 Infections Are About To Occur: It's Time To Stay Home [crosspost] Well I don't know how many of you reading this have been following the spread of Coronavirus. In the line of providing another person who is noticing the smoke in the room, I wrote this last night and started a blog to do it. Tl;d...
c3cb8581-d0ab-465f-8199-5909e2a91053
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
But is it really in Rome? An investigation of the ROME model editing technique *Thanks to Andrei Alexandru, Joe Collman, Michael Einhorn, Kyle McDonell, Daniel Paleka, and Neel Nanda for feedback on drafts and/or conversations which led to useful insights for this work. In addition, thank you to both William Saunders ...
1bb30aea-1338-448f-84cd-507423edfd47
trentmkelly/LessWrong-43k
LessWrong
Meetup : SF Meetup: Stories Discussion article for the meetup : SF Meetup: Stories WHEN: 06 June 2016 06:15:00PM (-0700) WHERE: 1597 Howard St., San Francisco, CA We'll be meeting to tell stories about our lives. Last time we had a list of prompts which helped a lot, so I'll be bringing those again. As usual, we e...
2efc643a-738a-40ec-bf4c-ab3f47b61617
trentmkelly/LessWrong-43k
LessWrong
Meetup : Washington DC: Jobs/project show and tell Discussion article for the meetup : Washington DC: Jobs/project show and tell WHEN: 13 October 2013 03:00:00PM (-0400) WHERE: Ballston Common mall We'll be meeting to tell each other about the things we're working on. (I think this would segue well into doing anoth...
e32ce475-5b71-4279-91ca-d99bcf5a16ff
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
An unaligned benchmark My goal is to design AI systems that are aligned with human interests and [competitive](https://ai-alignment.com/directions-and-desiderata-for-ai-control-b60fca0da8f4) with unaligned AI. I find it useful to have a particular AI algorithm in mind. Then I can think about how that algorithm could ...
2cd9701f-9328-4e0a-88ed-6467d08d845a
trentmkelly/LessWrong-43k
LessWrong
Drawing Less Wrong: Observing Reality To draw a city, you must walk around that city and look at it. You can't sit in a room with the blinds closed and create a map and expect it to be accurate. You cannot draw what you cannot see. To draw things, you need to look a things. This is surprisingly hard for a few reasons...
0dd497e8-375d-4d44-a25c-a3ae9f28bea5
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Shapes of Mind and Pluralism in Alignment *This post is part of the work done at* [*Conjecture*](https://conjecture.dev)*.* *This post has been written for the first* [*Refine*](https://www.alignmentforum.org/posts/5uiQkyKdejX3aEHLM/how-to-diversify-conceptual-alignment-the-model-behind) *blog post day, at the end of...
ce42226c-d0a7-4e16-977f-13a5477231ee
trentmkelly/LessWrong-43k
LessWrong
NSFW Content on LW There's some discussion of this on my post but I think it properly belongs on a meta post to which the mods can refer. Points for consideration: * Do we want content about sex on LW at all? Yes, but only on people's own blogs? Yes, but only with a content warning? Yes, but only with some sort of ...
71f0f85b-eb83-4279-87cf-9b8f086f25ac
trentmkelly/LessWrong-43k
LessWrong
Slide deck: Introduction to AI Safety This is from a presentation that I've been giving over the past few months to various groups within the Johns Hopkins University Applied Physics Laboratory (APL), where I work. I'd be interested to hear people's comments, critiques, suggestions, etc. Note that this is a somewhat ...
d74a6db9-4e74-4d4b-b238-c253c6d167d8
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
My take on higher-order game theory This is how I currently think about [higher-order game theory](https://www.alignmentforum.org/posts/FPML8k4QtjJxk3Y4M/confusions-re-higher-level-game-theory), the study of agents thinking about agents thinking about agents.... This post doesn't add any new big ideas beyond what wa...
b651830d-f571-4bb3-b8c0-29b5504b6b45
trentmkelly/LessWrong-43k
LessWrong
Lamentations, Gaza and Empathy Lamentations Today is the fast of Av, when Jews mourn the destruction of their temples in Jerusalem and their exile. We fast, sit on low stools, and read the book of lamentations (Eichah). This book, traditionally ascribed to the prophet Jeremiah, consists of 5 poems mourning the destru...
0f8e7d61-a92e-4517-bd70-b879bf59d073
trentmkelly/LessWrong-43k
LessWrong
An Open Letter To The Congregation Regarding The Upcoming Holiday Under normal circumstances, in a few days we would gather together to celebrate the festival of Purim. We celebrate by reading together the megillah. It tells a timeless story for every generation about standing up and avoiding death. Since, as we all ...
d6e51192-4484-4355-897b-ed7d134ed794
trentmkelly/LessWrong-43k
LessWrong
What and Why: Developmental Interpretability of Reinforcement Learning Introduction I happen to be in that happy stage in the research cycle where I ask for money so I can continue to work on things I think are important. Part of that means justifying what I want to work on to the satisfaction of the people who provid...
527c5852-5e6b-477d-a44f-9d20ea7a19a7
StampyAI/alignment-research-dataset/arxiv
Arxiv
An Optimistic Perspective on Offline Reinforcement Learning 1 Introduction --------------- One of the main reasons behind the success of deep learning is the availability of large and diverse datasets such as ImageNet (Deng et al., [2009](#bib.bib12)) to train expressive deep neural networks. By contrast, most rein...
09269fcc-263d-4984-9916-c37d70cbfaee
trentmkelly/LessWrong-43k
LessWrong
Adversarial Robustness Could Help Prevent Catastrophic Misuse There have been several discussions about the importance of adversarial robustness for scalable oversight. I’d like to point out that adversarial robustness is also important under a different threat model: catastrophic misuse.  For a brief summary of the ...
99569c30-64bd-42bf-94ae-f82f189bb2da
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
A Map to Navigate AI Governance Summary: -------- ![](https://39669.cdn.cke-cs.com/cgyAlfpLFBBiEjoXacnz/images/da871e6dbf234c1d8c24c0734dcd22c707cf2ddf8fdd1683.png)[**Summary figure - click here for full-size & editable file (google slide).**](https://docs.google.com/presentation/d/1SAArUwef9olVN-AVMnEBxQLgKhjgCkYsL_...
0a0c6354-0c77-4019-bb04-3419bbdcf6ef
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Apply to a small iteration of MLAB to be run in Oxford TLDR: We’re running a small iteration of MLAB (~10 participants) in Oxford towards the end of September. If you’re interested in participating, apply [here](https://forms.gle/QV9z6cxNGnS1UJzq6) by 7 September. If you’re interested in being a TA, please email us di...
e9267449-2d32-4a25-b112-6eb691c97e92
trentmkelly/LessWrong-43k
LessWrong
Austin, TX LW Meetup, 6/4 1:30PM After a brief hiatus, the Austin LW meetup is back together and stepping things up: for the foreseeable future, we plan on meeting every week. The next meeting is: Saturday 6/4, 1:30 PM, Caffé Medici; right across the street from UT. Parking is available on the street on in one of the...
532099c5-f19b-4314-94cf-2a9c34696b0f
trentmkelly/LessWrong-43k
LessWrong
Steering Language Models in Multiple Directions Simultaneously Narmeen developed, ideated and validated K-steering at Martian. Luke generated the baselines, figures and wrote this blog post. Amir proposed the research direction and supervised the project. The full interactive blog will be available closer to the publi...
427fe812-6741-4c16-867d-f3a6d9cab3e7
trentmkelly/LessWrong-43k
LessWrong
The Revelation Today the life of Alexander Kruel ends, or what he thought to be his life. He becomes aware that his life so far has been taking place in a virtual reality to nurture him. He now reached a point of mental stability that enables him to cope with the truth, hence it is finally revealed to him that he is a...
9afc5eb6-c974-49fa-b8f0-b7b95fa87c45
trentmkelly/LessWrong-43k
LessWrong
Anthropic reasoning and correlated decision making There seems to be some confusion on how to deal with correlated decision making - such as with absent-minded drivers and multiple copies of yourself; any situation in which many agents will allreach the same decision. Building on Nick Bostrom's division-of-responsibil...
e2ab0b54-bc8a-43eb-9954-aa6088ad32d9
trentmkelly/LessWrong-43k
LessWrong
Crisis Boot Camp: lessons learned and implications for EA Over the last 7+ months at work, I've needed to handle or support several crises (I'm on the US EV board and, in normal times, head of the community health and special projects team). It's been a crisis-handling boot camp, so I want to share my lessons learned....
3270e6cf-c944-4464-ab9b-eebd91ede416
trentmkelly/LessWrong-43k
LessWrong
Wobbly Table Theorem in Practice It turns out the “wobbly table theorem” is less of just a math curiosity than I thought. The “wobbly table theorem” essentially says that if you have a “table”[1] resting on a continuous surface, you can find a rotation where the table rests on all four legs[2]. I first encountered th...
97047fb8-f121-4083-8d2d-d2d204718957
trentmkelly/LessWrong-43k
LessWrong
Modified Demski Prior Problem: Revision The argument in A Problem with the Modified Demski Prior does not work as stated. The argument which I gave seems to make sense in the case where the random bits are added as axioms, but fails in the more natural case where we update on them in a Bayesian manner. To re-articula...
36ae0111-6e8a-4e58-bc4f-8032257b972e
trentmkelly/LessWrong-43k
LessWrong
Frequentist vs Bayesian breakdown: interpretation vs inference Suppose we have two different human beings, Connor and Diane, who agree to interpret their subjective anticipations as probabilities, thereby commonly earning them the title "Bayesian".  On a particular project or venture, they might disagree on Trick A or...
41e9774a-0838-4bbf-9729-725bcc90fd29
trentmkelly/LessWrong-43k
LessWrong
Russian x-risks newsletter spring 21 Despite vaccination with Russian Sputnik, I’ve got covid. It was relatively mild, but with elevated levels of D-dimer, which means higher risk of thrombosis. I am fine now. Notwithstanding availability of vaccines, only 10 per cent of Russian population has vaccinated and the reas...
c11a1889-d234-4141-8e4a-01fe8c49ffad
trentmkelly/LessWrong-43k
LessWrong
Distinguishing misuse is difficult and uncomfortable USPS delivers ibuprofen and vicodin just the same because it doesn't care enough to open and test every bottle of pills. If they did carefully inspect the contents of every package, they still couldn't distinguish other "shipping misuse" — money for your granddaught...
4978e34d-a4b4-4349-b2b3-14becc8cc310
trentmkelly/LessWrong-43k
LessWrong
Meetup : Brussels meetup Discussion article for the meetup : Brussels meetup WHEN: 20 April 2013 01:00:00PM (+0200) WHERE: Rue des Alexiens 55 1000 Bruxelles We are once again meeting at 'la fleur en papier dorée' close to the Brussels Central station. If you feel like an intelligent discussion and are in the neigh...
9883eb6b-ce7b-4ec8-80fd-1434d8df39c4
StampyAI/alignment-research-dataset/blogs
Blogs
New paper: “Functional Decision Theory” [![Functional Decision Theory](https://intelligence.org/files/fdt.png)](https://arxiv.org/abs/1710.05060) MIRI senior researcher Eliezer Yudkowsky and executive director Nate Soares have a new introductory paper out on decision theory: “**[Functional decision theory: A new theor...
b8ed6197-5a33-422b-b261-19b59c7c7b7b
StampyAI/alignment-research-dataset/arxiv
Arxiv
Multi-task Maximum Entropy Inverse Reinforcement Learning. 1 Introduction --------------- ![Multi-task inverse reinforcement learning (IRL) uses demonstrations of similar tasks, such as grasping different types of containers, to jointly infer the reward functions for each task. By exploiting the similarity between ...
5852131d-34fe-41e5-a4d2-b871153b67fd
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
GPT-4 is out: thread (& links) [GPT-4 is out](https://openai.com/research/gpt-4). There's also [a LessWrong post on this](https://www.lesswrong.com/posts/pckLdSgYWJ38NBFf8/gpt-4) with ~~some~~ a lot of discussion. The developers are doing [a live-stream ~~~now~~](https://www.lesswrong.com/posts/pckLdSgYWJ38NBFf8/gpt-4...
bd49c9b0-a962-40fc-9952-2d058011fb12
trentmkelly/LessWrong-43k
LessWrong
Research Motivations and "The Division of Cognitive Labor" Epistemic status: I’m moderately confident in the positions I endorse here, and this series is the product of several months’ research, but it only skims the surface of the literature on these questions. Bookkeeping: This post is part of a short series review...
9adeb9f8-9711-4ec0-8e95-0103510b6f14
trentmkelly/LessWrong-43k
LessWrong
Meetup : Seattle: Decision Theory Discussion article for the meetup : Seattle: Decision Theory WHEN: 01 April 2012 04:00:00PM (-0700) WHERE: 5523 University Way NE #501 Seattle, WA 98115 It's past time we had another meetup. As I mentioned earlier, it's past time for another meetup and, I'd like to have a discussio...
a157d120-f1a5-4459-82b6-c4a84225084e
trentmkelly/LessWrong-43k
LessWrong
Proposal on AI evaluation: false-proving Currently, one of arguments against deploying powerful non-agentic AIs is that they can deceive you: you ask for plan for doing something, AI creates one and then unexpectedly gets power as result of that plan. We can ask AI to provide proof or reasoning why its plan works. Ho...
d8a03322-fbfa-4146-a27f-b2c0db597b28
trentmkelly/LessWrong-43k
LessWrong
The role of epistemic vs. aleatory uncertainty in quantifying AI-Xrisk (Optional) Background: what are epistemic/aleatory uncertainty? Epistemic uncertainty is uncertainty about which model of a phenomenon is correct. It can be reduced by learning more about how things work. An example is distinguishing between a fair...
957e2aba-536a-4093-bbe7-0bd58b82e100
trentmkelly/LessWrong-43k
LessWrong
The 99% principle for personal problems Often when people are dealing with an issue – emotional, mental, or physical – there’s genuine progress and the issue becomes less and seems to go away. Until it comes back, seemingly as bad as before. Maybe the person developed a coping mechanism that worked, but only under s...
e7922421-d56d-4838-82b4-55797b5a7b1f
trentmkelly/LessWrong-43k
LessWrong
A method of writing content easily with little anxiety Hello, I have today been informed about a method for writing content that is relatively easy and painless. The idea is that by default a human will when reading predict what word is going to come next. Studies show that humans can predict what letter is going to c...
2627f2b9-7826-4c54-b69e-b54645a4f922
trentmkelly/LessWrong-43k
LessWrong
[U.S. Specific] Free money (~$5k-$30k) for Independent Contractors and grant recipients from U.S. government [crossposted from the EA Forum] Epistemic Status: I spent the past couple of days looking into this, but there is far more that I don’t know than that I do. I may be wrong about some of this, and this certainl...
0d3ea9ae-1703-4108-8fff-59c34295f461
trentmkelly/LessWrong-43k
LessWrong
Speedrunning 4 mistakes you make when your alignment strategy is based on formal proof Lots of people think formal (i.e. computer verified) proofs are a part of viable alignment strategies. This trick never works. If you think it might work, I can save your time in four different ways. I previously wrote on this, but...
e713b44a-5e60-475a-afd1-589b9925c3e7
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Is Democracy a Fad? *This cross-post from my* [*personal blog*](https://benmgarfinkel.wordpress.com/2021/02/26/is-democracy-a-fad/) *explains why I think democracy will probably recede within the next several centuries, supposing people are still around.* *The key points are that: (1) Up until the past couple centur...
a855c963-a36b-4ff8-abe2-acd54156e310
trentmkelly/LessWrong-43k
LessWrong
ChatGPT deceives users that it's cleared its memory when it hasn't Epistemic status: this is not fullproof evidence. Given that scheming in the wild is at least a little likely already, this looks like scheming. Others have offered hallucination-based explanations. What happened This is a real example from my ChatGP...
28e3cfc2-14b3-41a6-b199-44d7ffdc59b0
trentmkelly/LessWrong-43k
LessWrong
On the limits of idealized values (Cross-posted from Hands and Cities) On a popular view about meta-ethics, what you should value is determined by what an idealized version of you would value. Call this view “idealizing subjectivism.”  Idealizing subjectivism has been something like my best-guess meta-ethics. And lo...
e4474d16-650a-4b49-b91a-ed10d52ad1b4
trentmkelly/LessWrong-43k
LessWrong
Trustworthy and untrustworthy models In this post, I’ll talk about trustworthiness, a concept similar to scheming and deceptive alignment. I prefer this concept, as I find it A) is directly important in practice, B) more readily distinguishes between capability for scheming and in fact scheming, and C) is clearer how ...
a32c9a8d-6655-4995-9894-4c4d5964615f
StampyAI/alignment-research-dataset/arxiv
Arxiv
Empowerment for Continuous Agent-Environment Systems 1 Introduction --------------- One goal of AI research is to enable artificial agents (either virtual or physical ones) to act “intelligently” in complex and difficult environments. A common view is that intelligent behavior can be “engineered”; either by fully h...
1ec6cc2d-1c89-40ec-8075-79336b5b6770
trentmkelly/LessWrong-43k
LessWrong
Omicron Post #14 Important note on testing that I’m including at top of both posts today. From various sources, I have become convinced that rapid tests taken from nose swabs are likely to often be several days slower at detecting infections than rapid tests that use throat swabs. Here’s Washington Post on that, thi...
5a20393c-3684-4ed2-96c1-9d03298d1a3a
StampyAI/alignment-research-dataset/blogs
Blogs
Hamming questions and bottlenecks The [CFAR alumni workshop](https://www.eventbrite.com/e/the-hamming-question-boston-edition-tickets-16449605212) on the first weekend of May was focused on the Hamming question. Mathematician Richard Hamming was known to approach experts from other fields and ask “what are the importa...
f27125b4-14f1-4a9a-bb7e-57e6b7fae787
StampyAI/alignment-research-dataset/agisf
AGI Safety Fund
Least-to-Most Prompting Enables Complex Reasoning in Large Language Models 1 Introduction --------------- Despite the great success of deep learning in the past decade, there still remain huge differences between human intelligence and machine learning: (1) Given a new task, humans usually can learn to accomplish i...
38269e2a-d9e1-4a45-ad77-4554ea4e295d
StampyAI/alignment-research-dataset/agisf
AGI Safety Fund
Careers in alignment Part 1: Excerpt from Gleave’s [Careers in Beneficial AI Research](https://www.google.com/url?q=https://docs.google.com/document/d/1RFo7_9JVmt0z8RPwUjB-mUMgCMoUQmsaj2CM5aHvxCw/edit%23heading%3Dh.g5n8044splxv&sa=D&source=editors&ust=1692801594601440&usg=AOvVaw0xSu-rTZdoGrEwgnE6hupi) ----------------...
8d006843-61c8-4b49-bf38-3993d1da78eb
StampyAI/alignment-research-dataset/lesswrong
LessWrong
An explanation of decision theories [Epistemic status: I may have gotten some things wrong so please point out any errors. I also papered over a lot of technical details in favor of presenting my intuitions. A lot of this is me thinking out loud. If you already know a lot about decision theory or want the gory details...
57240790-f3fd-4304-a276-b6a02446773f
trentmkelly/LessWrong-43k
LessWrong
Alcoholics Anonymous: Much More Than You Wanted To Know [EDIT 10/27: Slight changes in response to feedback; correcting some definitions. I am not an expert in this field and will continue to make changes as I learn about them. There is a critique of this post here and other worse critiques elsewhere. My only excuse f...
0d0745c4-a398-4f48-85b7-60932ae35c54
trentmkelly/LessWrong-43k
LessWrong
Noticing and Overcoming Bias Introduction There’s a classic optical illusion about two squares on a chessboard. Square B looks significantly lighter, but they’re really the same shade - my mind is automatically correcting for the perceived shadow. I don’t think this particular quirk of the eye is that interesting. Bu...