id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
9d7bc8b3-c576-41dd-90ac-f1f1f934974f | trentmkelly/LessWrong-43k | LessWrong | Evidence on recursive self-improvement from current ML
The goal of this post is to start a collaborative analysis. Please post papers or links or other evidence I have missed in comments. I want to get some of Lesswrong's collective mind focused on this question.
A core component of the classical case for AI risk is... |
36b2be56-f3f5-4f60-a8a6-137e158a3b9a | trentmkelly/LessWrong-43k | LessWrong | Confusion about Normative Morality
Hi everyone,
If this has been covered before, I apologize for the clutter and ask to be redirected to the appropriate article or post.
I am increasingly confused about normative theories. I've read both Eliezer's and Luke's meta ethics sequences as well as some of nyan's posts, but... |
f87a343b-c14a-47d0-91e2-e31c3ea724ee | trentmkelly/LessWrong-43k | LessWrong | Covid 2/11: As Expected
Nothing both unexpected and important happened this week, other than that we might be on the verge of at least partially solving obesity.
There’s plenty else to talk about, but everything was on trend. That doesn’t mean my predictions were perfect or the speed of trends didn’t give us a few su... |
621a7785-25d2-48b8-97aa-84d0bd4dce2f | trentmkelly/LessWrong-43k | LessWrong | Junto: Questions for Meetups and Rando Convos
I ponder a lot about community and how important local community is for the functioning of society; many are the riches brought from afar by long distance communication. Nonetheless, local rationality meetups can increase local metis by generating intelligent community. I ... |
17a3cedf-de20-4090-a0bb-d4ee7d3b9c74 | trentmkelly/LessWrong-43k | LessWrong | Incentive design and capability elicitation
Introduction and summary
(This is the fourth in a series of four posts about how we might solve the alignment problem. See the first post for an introduction to this project and a summary of the series as a whole.)
In the first post in this series, I distinguished between “... |
f872701b-b07d-4336-a180-9ec32863c50d | trentmkelly/LessWrong-43k | LessWrong | Links for May
In English
What would happen if a superintelligent AI was aligned with your values?
The details here are a little too much in the “superintelligence is magic that can achieve anything” direction to my taste (I don’t think that anything will just be instantly teleported into safety, superintelligent AI ... |
70cb2ac4-f41a-42fa-b40f-d00a0bf0f5ca | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Notes on Meta's Diplomacy-Playing AI
*Disclaimer: I'm not an expert at machine learning, AI safety or Diplomacy, so there may be errors here, though hopefully no major ones. For previou... |
897b6d22-c564-416a-8c40-441dbe604dcc | StampyAI/alignment-research-dataset/lesswrong | LessWrong | What are the biggest current impacts of AI?
I'd like to have a clearer picture of the domains in which AI systems have already been deployed - particularly those in which they are having the largest impacts on the world, and what those impacts are.
Some reasons why this might be useful:
* AI systems don't (current... |
1813b3f6-af8e-48fd-a483-cb2b2366a55a | trentmkelly/LessWrong-43k | LessWrong | Punishing the good
Should you punish people for wronging others, or for making the wrong call about wronging others?
For example:
1. A newspaper sends me annoying emails all the time, but suppose that empirically if they didn’t behave like this, they would get markedly fewer subscribers, and may not survive. And su... |
6848206d-c379-4f63-81e4-a101a0cbb9c7 | trentmkelly/LessWrong-43k | LessWrong | [SEQ RERUN] Two Visions of Heritage
Today's post, Two Visions Of Heritage was originally published on December 9, 2008. A summary:
> Differences between Hanson and Yudkowsky on how they see heritage.
Discuss the post here (rather than in the comments to the original post).
This post is part of the Rerunning the... |
affc188a-7ca4-4522-9f8e-30156b1d3bc0 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Should the EA community have a DL engineering fellowship?
It seems that it is generally useful for people aiming to get a career in AI Safety to have some capability in ML software engineering. Some of the most straightforward ways I've come across are:
* fast.ai-like or Coursera courses.
* Bootcamps, such as the on... |
af5d41c1-8a4c-40cd-acea-fd8af93d2577 | trentmkelly/LessWrong-43k | LessWrong | EA & LW Forum Weekly Summary (13th - 19th March 2023)
This is part of a weekly series summarizing the top posts on the EA and LW forums - you can see the full collection here. The first post includes some details on purpose and methodology. Feedback, thoughts, and corrections are welcomed.
If you'd like to receive th... |
1bff96d9-d4fa-4da1-8fa4-08d316cb9d0b | StampyAI/alignment-research-dataset/arxiv | Arxiv | Regulatory Markets for AI Safety
1 Introduction
---------------
Ensuring that humans benefit from AI will depend on our capacity to regulate how it is built and deployed. There are risks arising from misuse of powerful technology and from misalignment between the goals an AI system is pursuing and the true goals of... |
85de31e2-3489-47b9-a50b-92a5a403d34b | trentmkelly/LessWrong-43k | LessWrong | Undergrad AI Safety Conference
TL;DR: undergrad AI safety conference in Chicago on world-modelling & thinking about the cruxes of the future of TAI. Takes place March 29-30, apply by Feb 20.
The UChicago AI Safety Team at XLab is excited to announce the Chicago Symposium on Transformative AI, an undergraduate AI Saf... |
2a191878-81d9-48e5-a360-3a3f5c532e22 | trentmkelly/LessWrong-43k | LessWrong | Computation Hazards
THIS IS A SUMMARY OF MATERIAL FROM VARIOUS POSTS AND DISCUSSIONS. MY THANKS TO ELIEZER YUDKOWSKY, DANIEL DEWEY, PAUL CHRISTIANO, NICK BECKSTEAD, AND SEVERAL OTHERS.
Several ideas have been floating around LessWrong that can be organized under one concept, relating to a subset of AI safety problems... |
067328e5-30e7-4d3c-b2ce-d7c8200c1936 | trentmkelly/LessWrong-43k | LessWrong | Should you write under a blog or your own name?
I've always wondered if it's better to write at a blog (overcomingbias.com) or write under your own name (scottaaronson.blog).
The benefits of writing under your own name seem to be easier credit to your work and that people get an instant snapshot of what you're intere... |
50b36c97-191a-4c43-8fcc-68edf3fcd165 | trentmkelly/LessWrong-43k | LessWrong | Morally underdefined situations can be deadly
Thanks to Rebecca Gorman for help with this post.
Morally underdefined situations
I recently argued that full understanding of value extrapolation[1] was necessary and almost sufficient for solving the AI alignment problem.
In it, I introduced situations beyond the huma... |
987017fb-5b14-409c-a6c4-136dd9d57775 | trentmkelly/LessWrong-43k | LessWrong | Industrial literacy
I’ve said before that understanding where our modern standard of living comes from, at a basic level, is a responsibility of every citizen in an industrial civilization. Let’s call it “industrial literacy.”
Industrial literacy is understanding…
* That the food you eat is grown using synthetic fe... |
297a63ee-f6b6-49d6-a1bf-5c9342ea3887 | trentmkelly/LessWrong-43k | LessWrong | New LW Meetup: Basel
This summary was posted to LW Main on March 6th. The following week's summary is here.
New meetups (or meetups with a hiatus of more than a year) are happening in:
* [Basel] Effective Altruism Meetup Switzerland: 21 March 2015 06:00PM
Irregularly scheduled Less Wrong meetups are taking place i... |
e7104211-f101-4bf1-aa2d-2fd5d5ac84cc | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Swap and Scale
*Produced as part of the SERI MATS program under John Wentworth*
*Thanks to Garrett Barker for pointing out you could think of Swap Symmetry as relabeling and to Matthias G. Mayer for catching my mistake in not multiplying the bias in Scale Symmetry.*
Here are two ways you can change the weights o... |
97b4ad8b-4475-4d6f-92ae-420ba6771955 | trentmkelly/LessWrong-43k | LessWrong | On Automoderation
Follow Up to: Trio Walks, Duo Talks
Response to: Automoderation (Ferocious Truth)
Hat Tip: Automoderation (Agenty Duck)
In my post Trio Walks, Duo Talks I laid out the thesis that the key to a high-quality discussion with more than two participants is to organize the group such that it takes on ma... |
af7dba17-d068-439f-94d0-673d801559b2 | trentmkelly/LessWrong-43k | LessWrong | Are minimal circuits deceptive?
This post is part of research I did at OpenAI with mentoring and guidance from Paul Christiano.
One possible inner alignment technique is to structure your learning algorithm’s inductive biases such that it is far more likely for it to produce an aligned model than a misaligned model. ... |
33f96a03-b0eb-4fd4-b9ba-044a73480382 | trentmkelly/LessWrong-43k | LessWrong | Deviant argumentation regarding Monty Hall problem
In class number four of the online Artificial Intelligence course from Stanford, Peter Norvig explains "one of the most popular problems in the subject of probability theory, the Monty Hall problem". His presentation was exactly what I have always been taught, and I g... |
a8a42aaf-f703-49eb-929b-1a18b8b21e63 | trentmkelly/LessWrong-43k | LessWrong | How Low Should Fruit Hang Before We Pick It?
,
Even if we can measure how impactful an agent's actions are, how impactful do we let the agent be? This post uncovers a surprising fact: armed with just four numbers, we can set the impact level so that the agent chooses a reasonable, non-catastrophic plan on the first tr... |
1959bb2c-18f3-4d72-8ebf-8b7b6b803847 | trentmkelly/LessWrong-43k | LessWrong | The easy goal inference problem is still hard
Posted as part of the AI Alignment Forum sequence on Value Learning.
> Rohin’s note: In this post (original here), Paul Christiano analyzes the ambitious value learning approach. He considers a more general view of ambitious value learning where you infer preferences more... |
78e8e661-f1af-4c40-9be3-d1d20c1f8fa7 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | What could an AI-caused existential catastrophe actually look like?
*This article forms part of 80000 Hours's explanation of* [*risks from artificial intelligence*](https://80000hours.org/problem-profiles/artificial-intelligence/)*, and focuses on* how *an AI system could cause an existential catastrophe. Our full* [... |
11a5a02a-5094-4061-ae60-fbdcc9961853 | trentmkelly/LessWrong-43k | LessWrong | Set Theory Multiverse vs Mathematical Truth - Philosophical Discussion
I've been thinking about the set theory multiverse and its philosophical implications, particularly regarding mathematical truth. While I understand the pragmatic benefits of the multiverse view, I'm struggling with its philosophical implications.
... |
126af534-bd99-4b23-924b-834925915932 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | What's a Decomposable Alignment Topic?
What's an alignment topic where, if someone decomposed the overall task, a small group of smart people (like here on Lesswrong) could make conceptual progress? By "smart", assume they can [notice confusion](https://www.lesswrong.com/s/zpCiuR4T343j9WkcK), google, and program.
Be ... |
4def184d-f61f-44e0-9d34-2d5fd40783ea | StampyAI/alignment-research-dataset/arbital | Arbital | Consistency
A consistent [https://arbital.com/p/-theory](https://arbital.com/p/-theory) is one in which there are well-formed statements that you cannot prove from its axioms; or equivalently, that there is no $X$ such that $T\vdash X$ and $T\vdash \neg X$.
From the point of view of [https://arbital.com/p/-model_theo... |
d6927803-f601-472c-8ec9-1a5e07d67f55 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "The concept of "logical rudeness" (which I'm pretty sure I first found here, HT) is one that I should write more about, one of these days. One develops a sense of the flow of discourse, the give and take of argument. It's possible to do things that completely derail that flow of discourse without shouting or swearing.... |
1d4e14b5-a86b-498d-9dda-030d43b5005f | trentmkelly/LessWrong-43k | LessWrong | GPT-4 is easily controlled/exploited with tricky decision theoretic dilemmas.
TL;DR: GPT-4 can fall for every decision theoretic adversary and control mechanism discussed in this paper except for a couple which it fails to understand in the first place.
This post is based in part on the 2020 paper Achilles Heels f... |
8ff38e85-0349-4686-9069-a72eb0a548c9 | StampyAI/alignment-research-dataset/arbital | Arbital | Complexity theory
**Complexity theory** is the study of the efficiency of algorithms with respect to several metrics, usually time and memory usage. Complexity theorists aim to classify different problems into [classes of difficulty](https://arbital.com/p/complexity_class) and study the relations that hold between the... |
e0479b34-44d1-486d-8a14-ed573e7acbc0 | trentmkelly/LessWrong-43k | LessWrong | Strategies for coalitions in unit-sum games
I'm going to formalize some ideas related to my previous post about pursuing convergent instrumental goals without good priors and prove theorems about how much power a coalition can guarantee. The upshot is that, while non-majority coalitions can't guarantee controlling a n... |
c61a3c65-9f6f-4247-959c-6affbdc0f101 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Newcastle-upon-Tyne meetup, December
Discussion article for the meetup : Newcastle-upon-Tyne meetup, December
WHEN: 07 December 2013 12:00:00PM (+0000)
WHERE: Baltic centre for contemporary art
I have no idea how many LWers there are in the Northeast of England who would be interested in a meetup, but as N... |
83879518-957b-4864-9779-4e7e61559913 | trentmkelly/LessWrong-43k | LessWrong | [AN #60] A new AI challenge: Minecraft agents that assist human players in creative mode
Find all Alignment Newsletter resources here. In particular, you can sign up, or look through this spreadsheet of all summaries that have ever been in the newsletter. I'm always happy to hear feedback; you can send it to me by rep... |
e688dd4d-97ae-442a-90a0-c9af5791cc4b | trentmkelly/LessWrong-43k | LessWrong | AI #118: Claude Ascendant
The big news of this week was of course the release of Claude 4 Opus. I offered two review posts: One on safety and alignment, and one on mundane utility, and a bonus fun post on Google’s Veo 3.
I am once again defaulting to Claude for most of my LLM needs, although I often will also check o... |
6fa40043-3240-4383-9d6a-cf3cfdca0a2f | StampyAI/alignment-research-dataset/blogs | Blogs | Machine Intelligence Research Institute Progress Report, March 2012
Past progress reports: [February 2012](http://intelligence.org/blog/2012/03/03/singularity-institute-progress-report-february-2012/), [January 2012](http://intelligence.org/blog/2012/02/05/singularity-institute-progress-report-january-2012/), [Decembe... |
6e067ea3-3786-4adf-80e4-e4a13104c203 | trentmkelly/LessWrong-43k | LessWrong | Meetup : LW/Methods of Rationality meetup
Discussion article for the meetup : LW/Methods of Rationality meetup
WHEN: 18 October 2013 07:00:00PM (-0400)
WHERE: MIT building 6, room 120 Cambridge, MA
On Oct 18th at 7pm there will be a Less Wrong / Methods of Rationality meetup/party on the MIT campus in Building 6, r... |
fb1da332-5760-4ef7-85fe-01658c369245 | trentmkelly/LessWrong-43k | LessWrong | Fucking Goddamn Basics of Rationalist Discourse
1. Don't say false shit omg this one's so basic what are you even doing. And to be perfectly fucking clear "false shit" includes exaggeration for dramatic effect. Exaggeration is just another way for shit to be false.
2. You do NOT (necessarily) know what you fucking... |
8f8ef533-6520-44f8-b5ec-362c6f791ba3 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | [AN #92]: Learning good representations with contrastive predictive coding
**Newsletter #92**
Alignment Newsletter is a weekly publication with recent content relevant to AI alignment around the world. Find all Alignment Newsletter **[resources here](http://rohinshah.com/alignment-newsletter/)**. In particular, you c... |
8dd5922a-d08d-4113-a8ee-57f0f476a116 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Amsterdam
Discussion article for the meetup : Amsterdam
WHEN: 23 November 2013 02:00:00PM (+0100)
WHERE: Oosterdokseiland 4, Amsterdam, The Netherlands
Let's have a Netherlands LessWrong meetup on Saturday the 23rd. We're meeting in the Starbucks / "East Dock Lounge" on the Oosterdokseiland next to Amsterd... |
285260a5-5ca3-4da9-ba51-106370bb2780 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | AI Risk Intro 1: Advanced AI Might Be Very Bad
*This is a submission for the* [*AI Safety Public Materials contest.*](https://www.lesswrong.com/posts/gWM8cgZgZ9GQAYTqF/usd20k-in-bounties-for-ai-safety-public-materials#How_to_submit)
**Introduction**
================
If human civilisation is destroyed this century, t... |
2743ce0d-733d-4b24-a4a8-b68dbfdb272f | trentmkelly/LessWrong-43k | LessWrong | Why not spend more time looking at human alignment?
As we approach AGI, we also approach the risk of alignment failure: due to either mismatch between intended goals and specified goals ('outer misalignment'), or mismatch between specified goals and emergent goals ('inner misalignment'), we end up with a catastrophica... |
c9accad6-375f-47a4-8334-7af844c04b1e | trentmkelly/LessWrong-43k | LessWrong | Crossword puzzle: LessWrong Halloween 2022
Just in time for Halloween - the spookiest themed crossword puzzle of the year!
* Interactive web version
* Printable PDF
The comments below may contain unmarked spoilers. I don't expect many people will solve it without doing at least some Wikipedia lookups, but you're w... |
f7ea88ae-f698-4bfe-ae99-d58ea1b72a30 | trentmkelly/LessWrong-43k | LessWrong | Program Search and Incomplete Understanding
,,,,,,
epistemic status: crystallizing a pattern
When trying to solve some foundational AI problem, there is a series of steps on the road to implementation that goes something like this (and steps may be skipped):
First, there is developing some philisophical solution to ... |
ca48d1c7-31f1-4f14-84e4-611bd4ad6f15 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Deep Learning for Symbolic Mathematics
1 Introduction
---------------
A longstanding tradition in machine learning opposes rule-based inference to statistical learning (Rumelhart et al., [1986](#bib.bib16 "Parallel distributed processing: explorations in the microstructure of cognition, vol. 1: foundations")), and ... |
dc9b9328-5d8d-460e-b020-8ddb314725a2 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | What I mean by "alignment is in large part about making cognition aimable at all"
*(Epistemic status: attempting to clear up a misunderstanding about points I have attempted to make in the past. This post is not intended as an argument for those points.)*
I have long said that the lion's share of the AI alignment pro... |
6fb84e3d-b56e-4cb5-a198-c58629c85f26 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | No, EDT Did Not Get It Right All Along: Why the Coin Flip Creation Problem Is Irrelevant
Back in 2017, [Johannes\_Treutlein](https://www.lesswrong.com/users/johannes_treutlein) published a post critiquing logical decision theories: [Did EDT get it right all along? Introducing yet another medical Newcomb problem](https... |
3eff75c3-3cd5-482c-97f7-aeddf73873e2 | trentmkelly/LessWrong-43k | LessWrong | What steep learning curve do you wish you'd climbed sooner?
This is the question asked by John Cook on Twitter. He lists responses from different people:
> * R
> * Version control
> * Linear algebra
> * Advanced math
> * Bayesian statistics
> * Category theory
> * Foreign languages
> * How to not waste time
>... |
13e24e00-e367-4c33-947e-7c6800953541 | trentmkelly/LessWrong-43k | LessWrong | Do we get better or worse at adapting to change?
Verner Vinge, in a classic 1993 essay, described “the Singularity” as an era where progress becomes “an exponential runaway beyond any hope of control.”
The idea that technological change might accelerate to a pace faster than we can keep up with is a common concern. A... |
62e83445-7ebe-4e9c-9937-99dba45ce669 | trentmkelly/LessWrong-43k | LessWrong | Making a Kingfisher Video
Cecilia and I recently made a Kingfisher video, and I wanted to write some about the process. I'm not claiming this is the best way to do it, and really I'm pretty sure it isn't. If people have feedback on how to do this better or more easily, I'm definitely interested!
We started with my pa... |
7a066c7c-6f62-450d-8ae3-94275602388c | trentmkelly/LessWrong-43k | LessWrong | Chaos and Stability
Human lifespan, athletic performance and the number of children a woman can gestate are short-tailed. We operate within hard physical limits. Wealth, power, fame, pandemics, religions, pleasure and pain are long-tailed. They are Pareto-Distributed. Less than 1% of the world population owns more tha... |
61eafa96-ba97-4937-bba6-25fa422fe0cd | trentmkelly/LessWrong-43k | LessWrong | LessWrong could grow a lot, but we're doing it wrong.
How do I know this? I got a copy of the website analytics.
The bounce rate for LessWrong's home page is 60%!
To be clear: Over half the people who visit LessWrong are going away without even clicking anything.
Yet how many NEW visitors are there? Almost half o... |
09b01b66-55dd-436b-91eb-0b69d4b95470 | trentmkelly/LessWrong-43k | LessWrong | Universal Love Integration Test: Hitler
I'm still not satisfied with this post, but thought I'd ship it since I refer to the concept a fair amount. I write this more as "someone who feels some kernel of univeral-love-shaped thing", but, like, i dunno man i'm not a love expert.
tl;dr
I think "love" means "To care abo... |
59947fc6-223c-4a54-b12f-235140ba5f40 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Recursive Middle Manager Hell: AI Edition
Two-sentence summary
====================
In [Recursive Middle Manager Hell](https://www.lesswrong.com/posts/pHfPvb4JMhGDr4B7n/recursive-middle-manager-hell), Raemon argues that organisations get increasingly warped (immoral/maze-like/misaligned/...) as they gain more levels ... |
ff3e049e-1622-40cc-ac49-3003895aaeb8 | StampyAI/alignment-research-dataset/blogs | Blogs | EleutherAI Second Retrospective: The long version
We've been fairly busy at Eleuther for the past year-and-a-half, here's the full story.
July 2021 – December 2021: An Interlude[#](#july-2021--december-2021-an-interlude)
------------------------------------------------... |
1ebc05f5-f3de-402a-a0c0-3f26f52dc4f3 | trentmkelly/LessWrong-43k | LessWrong | [LINK] Open Source Software Developer with Terminal Illness Hopes to Opt Out of Death
Aaron Winborn writes:
> TLDR: http://venturist.info/aaron-winborn-charity.html
>
> So maybe you've heard about my plight, in which I wrestle Lou Gehrig in this losing battle to stay alive. And I use the phrase "staying alive" lo... |
f498916d-e496-425f-b214-3cd290a2181d | trentmkelly/LessWrong-43k | LessWrong | Mixed strategy Nash equilibrium
Inspired by: Swords and Armor: A Game Theory Thought Experiment
Recently, nick012000 has posted Swords and Armor: A Game Theory Thought Experiment. I was disappointed to see many confused replies to this post, even after a complete solution was given by Steve_Rayhawk. I thought someone... |
b3d22bfa-66a9-478d-b441-6093c1373d3c | trentmkelly/LessWrong-43k | LessWrong | Slate Star Codex and Silicon Valley’s War Against the Media
None |
5b83c01b-5e98-4948-b4de-05799b53ab0c | trentmkelly/LessWrong-43k | LessWrong | Frontier Models are Capable of In-context Scheming
This is a brief summary of what we believe to be the most important takeaways from our new paper and from our findings shown in the o1 system card. We also specifically clarify what we think we did NOT show.
Paper: https://www.apolloresearch.ai/research/scheming-rea... |
d7886aab-0712-4ec8-bf0c-3bbce2264720 | trentmkelly/LessWrong-43k | LessWrong | Counterfactual Calculation and Observational Knowledge
Consider the following thought experiment ("Counterfactual Calculation"):
> You are taking a test, which includes a question: "Is Q an even number?", where Q is a complicated formula that resolves to some natural number. There is no a priori reason for you to exp... |
f995bd88-31ac-4861-a5bc-9b4d597e8e1b | trentmkelly/LessWrong-43k | LessWrong | January 2014 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... |
7331716f-54f0-4d7f-9647-fbef6b1a729b | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Inner Alignment in Salt-Starved Rats
*(See comment* [*here*](https://www.lesswrong.com/posts/wcNEXDHowiWkRxDNv/inner-alignment-in-salt-starved-rats?commentId=Lm5GwMqFnavJN2kHq#Lm5GwMqFnavJN2kHq) *for some corrections and retractions. —Steve, 2022)*
Introduction: The Dead Sea Salt Experiment
==========================... |
e394476c-af4e-4176-addf-0a6a57e3e2a7 | trentmkelly/LessWrong-43k | LessWrong | Is corruption a valuable antidote to overregulation?
When reading Robert Caro's book about Robert Moses there's a part of my that things that I'm glad that the city in which I'm living is less corrupt then New York was when Robert Moses was active.
On the other hand Robert Moses got things done and build a lot of in... |
88136f7b-8895-4aae-b5bd-682f40d729f2 | trentmkelly/LessWrong-43k | LessWrong | Weekly LW Meetups
This summary was posted to LW Main on May 6th. The following week's summary is here.
Irregularly scheduled Less Wrong meetups are taking place in:
* European Community Weekend: 02 September 2016 03:35PM
The remaining meetups take place in cities with regular scheduling, but involve a change in t... |
de550d1e-80a4-43e6-bdce-823c83c6ad67 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | AXRP Episode 20 - ‘Reform’ AI Alignment with Scott Aaronson
[Google Podcasts link](https://podcasts.google.com/feed/aHR0cHM6Ly9heHJwb2RjYXN0LmxpYnN5bi5jb20vcnNz/episode/NTM1YTQ1MzAtMDVlNS00OTE4LThlMjgtOWRmZWUzMjM1Mjk0)
How should we scientifically think about the impact of AI on human civilization, and whether or no... |
fc61d1c2-59ef-4ee9-baf7-2b22fe0b5c1f | trentmkelly/LessWrong-43k | LessWrong | Forcing yourself to keep your identity small is self-harm
The title lays out my thesis in perhaps the maximally provocative way: if you put direct effort into keeping your identity small (à la Graham), you are doing harm to yourself akin to the kind of harm you do to yourself by suppressing emotions.
I was inspired t... |
c7c54a1f-fe49-4a46-97aa-41760d56df57 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post87
UPDATE: Recent work with improved AF and compliance gap classifiers disagrees with our results. We recommend using the improved classifiers. Summary We wanted to briefly share an early takeaway from our exploration into alignment faking: the phenomenon appears fairly rare among the smaller open-source model... |
c84fa075-804e-4477-be15-10a941afc4d8 | trentmkelly/LessWrong-43k | LessWrong | List of most interesting ideas I encountered in my life, ranked
1. Bayesian thinking
It litterally was n°1 in my list, I was really happy to find this website, will not detail.
2. Active ignorance/avoidance and selective attention/participation
Instead of thinking, commenting, or saying something is stupid/b... |
912411fc-3f76-46a5-9e71-b975db7ecb25 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: 8794b0bd106aa54a
arXiv:2505.01420v4 [cs.LG] 3 Jul 2025
2025-7-4
Evaluating Frontier Models for Stealth and
Situational Awareness
Mary Phuong*,1, Roland S. Zimmermann*,1, Ziyue Wang*,1, David Lindner1, Victoria Krakovna1, Sarah
Cogan1, Allan Dafoe1, Lewis Ho1and Rohin Shah1
*Equal contribution,1Google DeepMind
Re... |
bdab311b-a5a9-457c-8053-c87cc9a04841 | trentmkelly/LessWrong-43k | LessWrong | Deceptive agents can collude to hide dangerous features in SAEs
TL;DR: We use refusal-ablated Llama 3 70B agents as both labeling and simulating agents on the features of a sparse auto-encoder, with GPT-4 as an overseer. The agents follow a deceptive coordinated strategy to mislabel important features such as "coup" o... |
2ebf97cd-ca73-49ef-bcfe-c20719a05c68 | trentmkelly/LessWrong-43k | LessWrong | Thoughts on the AI Safety Summit company policy requests and responses
Over the next two days, the UK government is hosting an AI Safety Summit focused on “the safe and responsible development of frontier AI”. They requested that seven companies (Amazon, Anthropic, DeepMind, Inflection, Meta, Microsoft, and OpenAI) “o... |
57662b8a-d741-4b9e-a102-69bd6fed4609 | trentmkelly/LessWrong-43k | LessWrong | How to write a mathematical formula on the fear of death?
I was reading the Methods of Rationality, and I was reading the part about how it's irrational to fear death. Well I came across "All x: Die(x) = Not Exist x: Not Die(x)" I really don't get this.. I'm sorry, I'm not good at math. But does "x" here represent an ... |
d07c8290-8e43-4e6c-ab5b-02369ffdaf40 | trentmkelly/LessWrong-43k | LessWrong | The Witching Hour
On an ordinary evening, Tal Aivon was lively and pleasant. The collection of longhouses and yurts within its tall brick walls shone bright with kerosene – not just torches, real kerosene – and its communal meeting area was noisy with conversation and song. The children would be playing their games, a... |
908815ce-6ed8-46ca-8efd-008f77053404 | trentmkelly/LessWrong-43k | LessWrong | Forecasting for Policy (FORPOL) - Main takeaways, practical learnings & report
Tl;dr: In Czech Priorities, a think tank based in Prague, together with Metaculus, we recently ran a complex forecasting tournament funded by EAIF to support Czech policymaking. Today, we are publishing a write-up of our findings from the t... |
14f19883-fe35-4a02-b7bc-d7518cd1b3ad | trentmkelly/LessWrong-43k | LessWrong | The Doomsday argument in anthropic decision theory
EDIT: added a simplified version here.
Crossposted at the intelligent agents forum.
In Anthropic Decision Theory (ADT), behaviours that resemble the Self Sampling Assumption (SSA) derive from average utilitarian preferences (and from certain specific selfish prefere... |
a575ad52-eaa5-41fa-85b8-8df1b3771338 | trentmkelly/LessWrong-43k | LessWrong | Novelty Generation - The Art of Good Ideas
Recently I set out to understand how new ideas are formed in the mind. My process consisted of poking at my own mind, poking at other people's minds, and then organizing group sessions of poking at each other's minds.
The Scientific Process™
The following is a list of insi... |
0a3e6e2d-94fa-413c-b4d0-822f7643dda1 | trentmkelly/LessWrong-43k | LessWrong | A Critique of Leverage Research's Connection Theory
Leverage Research is a recently formed New York-based research group "dedicate[d] to making the world a better place by the best and most effective means possible". I personally empathize and connect with this mission statement a lot, and I myself have made more or l... |
020bbf4f-1875-4c8a-9b9f-735987692398 | trentmkelly/LessWrong-43k | LessWrong | Review: Slay the Spire
Epistemic Status: Many hours played
Spoiler-Free Bottom Line: Slay the Spire is an amazing single-player roguelike deckbuilding game. When I wrote that Artifact was the most fun I’ve had gaming in a long time, the only alternative to give me pause was Slay the Spire. Each game, you work your wa... |
e8642ef5-a65a-48cb-b929-e23e2935a8b7 | trentmkelly/LessWrong-43k | LessWrong | New circumstances, new values?
A putative new idea for AI control; index here.
Quick, is there anything wrong with a ten minute pleasant low-intensity conversation with someone we happen to disagree with?
Our moral intuitions say no, as do our legal system and most philosophical or political ideals since the enlight... |
c5980b24-102f-4613-a132-a4ab918b96a7 | trentmkelly/LessWrong-43k | LessWrong | Introduction to Prisoners' Dilemma
Related to: Previous posts on the Prisoners' Dilemma
Sometimes Nash equilibria just don't match our intuitive criteria for a good outcome. The classic example is the Prisoners' Dilemma.
The police arrest two criminals, Alice and Bob, on suspicion of murder. The police admit they do... |
706ae17c-f969-4b3b-8099-e0703a9daf95 | trentmkelly/LessWrong-43k | LessWrong | The Power Users We Forgot: Why AI Needs Them Now More Than Ever
The People Who Were Right Too Early
There’s a quiet frustration shared by many in this community.
You raised concerns about AI before it was cool.
You explored edge cases before they were dangerous.
You built frameworks to understand systems that didn’t ... |
0fa5618b-829c-44be-b526-817023d5ca0e | StampyAI/alignment-research-dataset/blogs | Blogs | April 2017 Newsletter
| |
| --- |
|
Our newest publication, “[Cheating Death in Damascus](https://intelligence.org/2017/03/18/new-paper-cheating-death-in-damascus/),” makes the case for functional decision theory, our general framework for thinking about rational choice and counterfactual reasoning.
In other news, ... |
d2a58f99-1b26-4387-a484-2a80025b8f25 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | GPT-4 for personal productivity: online distraction blocker
There are many apps for blocking distracting websites: [freedom.to,](https://freedom.to/) [leechblock](https://www.proginosko.com/leechblock/), [selfcontrol](https://selfcontrolapp.com/), [coldturkey](https://getcoldturkey.com/), just to name a few. They are ... |
1ab4a4b9-84a9-43a9-9c2b-3d3d5d6a0b6e | trentmkelly/LessWrong-43k | LessWrong | Three Months In, Evaluating Three Rationalist Cases for Trump
I recall seeing three “rationalist” cases for Trump:
1. Richard Ngo on Twitter and elsewhere focused on the realignment of elite coalitions, observing that “most elite institutions have become leftist monocultures” but speculating that “over the next 5–10... |
9d343466-df8f-4b51-a635-65bdca5cc2b8 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post3738
Related to: Shut up and do the impossible! ; Everything about an AI in a box. One solution to the problem of friendliness is to develop a self-improving, unfriendly AI, put it in a box, and ask it to make a friendly AI for us. This gets around the incredible difficulty of developing a friendly AI, but it... |
b6540a82-ae47-4a13-b3c4-317a15b4358d | trentmkelly/LessWrong-43k | LessWrong | Asymptotic Logical Uncertainty: Introduction
In this post, I will introduce a new way of thinking about logical uncertainty. The main goal of logical uncertainty is to learn how to assign probabilities to logical sentences which have not yet been proven true or false.
One common approach is to change the question, as... |
85c8f3ad-0dd4-473c-8161-6bbcf895ab19 | trentmkelly/LessWrong-43k | LessWrong | Interview with Putanumonit
If you’ve ever wondered what Putanumonit will sound like as a podcast, wonder no more! An interview with me was recorded at a public meetup, it started with the basic tenets of rationality and putting numbers on things, then seamlessly segued to audience questions about Israel-Palestine and ... |
03eea826-f18b-448a-9827-b213d07ace10 | trentmkelly/LessWrong-43k | LessWrong | TSR #10: Creative Processes
This is part of a series of posts where I call out some ideas from the latest edition of The Strategic Review (written by Sebastian Marshall), and give some prompts and questions that I think people might find useful to answer. I include a summary of the most recent edition, but it's not a ... |
9cca6cab-0a11-4144-bb66-abc0553cdc7a | trentmkelly/LessWrong-43k | LessWrong | Are our community grouphouses typically rented, or owned?
Asking partially because the Auckland community really needs to start a grouphouse and could do with advice, and also I expect the answers to be interesting for a bunch of other reasons:
* Intuitively, one might expect a mostly longtermist community to be int... |
c74a08a2-3e67-4f58-8d2c-cbfc581e44a3 | trentmkelly/LessWrong-43k | LessWrong | Is protecting yourself from your own biases self-defeating?
I graduated from high school and wish to further my education formally by studying for a bachelor's degree in order to become a medical researcher. I could, for instance, take two different academic paths:
1. Study Medicine at undergraduate level and then ... |
54f4f651-f6ea-4756-9e4e-ea478fef537f | trentmkelly/LessWrong-43k | LessWrong | Toy model #6: Rationality and partial preferences
In my research agenda on synthesising human preferences, I didn't mention explicitly using human rationality to sort through conflicting partial preferences.
This was, in practice, deferred to the "meta-preferences about synthesis". In this view, rationality is just o... |
0f2ffa1d-7dcf-42f8-b368-28c6b6c39b0f | StampyAI/alignment-research-dataset/arxiv | Arxiv | An AGI with Time-Inconsistent Preferences
An AGI with Time -Inconsistent Preferences
May 28, 2019
James D. Miller and Roman Yampolskiy
Jdmiller@Smith.edu , roman.yampolskiy@louisville. edu
Introduction
This paper reveals a trap for artificial general intelligence (AGI) theorists who use ... |
3a6b56f1-b3ca-43aa-a569-3febad1dc17b | trentmkelly/LessWrong-43k | LessWrong | What isn't the wiki for?
The new wiki is off to a flying start - it's less than 15 hours old, and already it has over 20 articles and five authors. It's probably about time we worked out what it was for.
I created it because as things stand, I can't point my friends to Less Wrong and say "come and learn about rationa... |
2f92080d-ed36-4046-bd6f-5079bc2de923 | trentmkelly/LessWrong-43k | LessWrong | The $100B plan with "70% risk of killing us all" w Stephen Fry [video]
A high production value 16-minute video that summarizes the popular safety concerns, featuring Hinton, Russell and Claude 3.5. |
dfbba942-419b-43e2-9661-2b127d8fc9b4 | trentmkelly/LessWrong-43k | LessWrong | Therapist in the Weights: Risks of Hyper-Introspection in Future AI Systems
Epistemic Status: Speculative, but grounded in current trends in interpretability and self-supervised learning. This was initially written as an outline at some point in 2024, but written/rewritten more fully in response to Anthropic’s announc... |
62e9f94a-fdcc-4b7f-bffe-c23ed8f01308 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | UDT shows that decision theory is more puzzling than ever
I feel like MIRI perhaps mispositioned FDT (their variant of UDT) as a clear advancement in decision theory, whereas maybe they could have attracted more attention/interest from academic philosophy if the framing was instead that the UDT line of thinking shows ... |
8de9f568-342f-4f31-a3eb-b025a6967e31 | trentmkelly/LessWrong-43k | LessWrong | Does anyone have the reference to the "researchers should be spending more time on literature reviews" study?
This has been driving me crazy: about a month or two ago, I started (and kept) seeing various references to a study that basically argued that in many/some fields, the incremental benefit of researching a thin... |
1cb4496a-fba4-41a1-9d78-efcf1f48397d | trentmkelly/LessWrong-43k | LessWrong | Review of "Why AI is Harder Than We Think"
"Why AI is Harder Than We Think" is a recent (April 26, 2021) arXiv preprint by Melanie Mitchell. While the author is a tenured professor with technical publications, they also have published multiple layman-level essays and articles which tend to be skeptical and pessimistic... |
18a585f0-8ae9-4052-9fac-01b08ce4b53b | trentmkelly/LessWrong-43k | LessWrong | Interview with Bill O’Rourke - Russian Corruption, Putin, Applied Ethics, and More
This is cross-posted from my blog and I interviewed him for a general audience rather than LessWrong people.
I was lucky enough to interview Bill O’Rourke.
Imagine you’re on a plane to Russia to become the new president of an aluminum... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.