id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
9a91ed4e-469e-45a4-b630-2b4896f16c09 | trentmkelly/LessWrong-43k | LessWrong | Changes in College Admissions
This post brings together various questions about the college application process, as well as practical considerations of where to apply and go. We are seeing some encouraging developments, but mostly the situation remains rather terrible for all concerned.
APPLICATION STRATEGY AND DIFFI... |
1ffd9db5-eab8-429d-91f0-5420e1d42bb2 | trentmkelly/LessWrong-43k | LessWrong | 2025 Color Trends
Once again, I’ve compiled some statistics on color trends in the spring/summer 2025 ready-to-wear fashion collections!
Background and Methodology
We just got done with “fashion month”, the flurry of activity in the fall when fashion designers release their spring/summer collections for the coming y... |
e516b8f8-61e5-4c46-9e69-5d651b477f9c | trentmkelly/LessWrong-43k | LessWrong | Altman blog on post-AGI world
First part just talks about scaling laws, nothing really new. Second part is apparently his latest thoughts on a post-AGI world. Key part:
> While we never want to be reckless and there will likely be some major decisions and limitations related to AGI safety that will be unpopular, dire... |
30858163-8f00-4475-8dea-6ec5ebdde0ae | trentmkelly/LessWrong-43k | LessWrong | [Secular Solstice UK] We all have a part to play
|
61431c92-76d6-4600-bba2-356b8d13d14c | StampyAI/alignment-research-dataset/blogs | Blogs | Future-proof ethics
*Click lower right to download or find on Apple Podcasts, Spotify, Stitcher, etc.*
Ethics based on "common sense" seems to have a horrible track record.
That is: simply going with our intuitions and societal norms has, in the past, meant endorsing all kinds of insanity. To quote an article b... |
97a09c4d-ecf1-4d8d-9212-19946c359224 | trentmkelly/LessWrong-43k | LessWrong | Why we should fear the Paperclipper [Link]
> The Scenario
> A programmer has constructed an artificial intelligence based on an architecture similar to Marcus Hutter's AIXI model (see below for a few details). This AI will maximize the reward given by a utility function the programmer has given it. Just as a test, he ... |
3d4e64a1-e318-4ee7-812d-048d4e48e22c | trentmkelly/LessWrong-43k | LessWrong | 4 generations of alignment
Today, when I was scrapping a sticker urging to "stop immigration!" from a bus station, my son asked me what was the sticker's message about, and I've said it was calling for hate, and he asked "What is 'hate'?". My wife had tears of joy in eyes, and for once I had this feeling that perhaps ... |
0973143f-aa24-4e8b-a986-21c696909fa6 | StampyAI/alignment-research-dataset/blogs | Blogs | hackable multiverse
hackable multiverse
-------------------
in [a previous post](brittle-physics.html) i talk about how hackable physics might allow a superintelligence to take over very quickly (perhaps faster than the speed of light).
in *[psi rewriting](psi-rewriting.html)* i propose that multiversehood can be ... |
20add21a-91b9-423c-8f4b-ae34d961d045 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Moscow 11 February meetup
Discussion article for the meetup : Moscow 11 February meetup
WHEN: 11 February 2012 06:00:00PM (+0400)
WHERE: Mayakovskaya (Subway Station), Moscow, Russia
Despite the map (I have no idea why, but it shows wrong place), location of the meetup is Metro Mayakovskaya (in the center ... |
4dc58f24-85ef-4b3a-9f2e-8162b7345ee9 | trentmkelly/LessWrong-43k | LessWrong | An impossibility result for doing without good priors
My previous post suggested a strategy for making bets on uncertain propositions (e.g. the color of the sun) without having a good prior. Here I present an impossibility result for doing this in general.
----------------------------------------
Say there are n age... |
f2491bd9-7fa7-48e7-9148-31a6e9e0ad97 | trentmkelly/LessWrong-43k | LessWrong | Take the EA survey, help the EA movement grow and potentially win $250 to your favorite charity
This year's EA Survey is now ready to be shared! This is a survey of all EAs to learn about the movement and how it can improve. The data collected in the survey is used to help EA groups improve and grow EA. Data is also u... |
6f95b587-6363-4a14-b33b-40ab94d20c4d | trentmkelly/LessWrong-43k | LessWrong | Self-Adapting Language Models (from MIT, arXiv preprint)
I am not affiliated with the authors, mainly posting this to get some technical commentary on it. Full arXiv paper here.
> Large language models (LLMs) are powerful but static; they lack mechanisms to adapt their weights in response to new tasks, knowledge, ... |
2879f019-4586-42c1-b1bf-72bc0e4aaee8 | trentmkelly/LessWrong-43k | LessWrong | Meta Alignment: Communication Guide
I’ve been working on a series of longer articles explaining and justifying the following principles of communicating AI risk to the general public, but in the interest of speed and brevity, I’ve decided to write a simplified guide here. Some of this will be based o... |
ee737c55-7cc3-42c5-b8b9-f95c555b475e | trentmkelly/LessWrong-43k | LessWrong | The Signal and the Corrective
This is one of the most important articles I've read in awhile. It makes a generalization from something Scott pointed out in this SSC post.
Here are a few excerpts, but it's worth clicking through and reading the whole thing!
> From the inside, when you subscribe to a narrative, when y... |
9c208f1d-214b-4a86-a848-37a4eb72392f | trentmkelly/LessWrong-43k | LessWrong | How do I better stick to a morning schedule?
I've always struggled with keeping my morning routine to a reasonable length of time. I'm spacey and unfocused in the morning, and it can be hard to stay aware of "when" I am in the morning.
I don't want to come down too hard on this part of the day, because it also feels ... |
1d7d811e-2389-455c-b957-1e2f45381239 | trentmkelly/LessWrong-43k | LessWrong | Come to Manifest 2024 (June 7-9 in Berkeley)
TLDR
Manifold is hosting a festival for prediction markets: Manifest 2024! We’ll have serious talks, attendee-run workshops, and fun side events over the weekend. Chat with special guests like Nate Silver, Scott Alexander, Robin Hanson, Dwarkesh Patel, Cate Hall, and more a... |
992bab0b-6bee-421c-9466-6a7355a514fd | StampyAI/alignment-research-dataset/arbital | Arbital | Distances between cognitive domains
In the context of [AI alignment](https://arbital.com/p/2v), we may care a lot about the degree to which competence in two different [cognitive domains](https://arbital.com/p/7vf) is separable, or alternatively highly tangled, relative to the class of algorithms reasoning about them.... |
0c8f75e2-8921-4728-8d31-de049fd229f0 | trentmkelly/LessWrong-43k | LessWrong | Self-Control Secrets of the Puritan Masters
People today are fighting the temptations of a new economy. With Twitter and TikTok a click away, we need self-control to get anything done. There’s an explosion of new movements and techniques to fight procrastination, from Pomodoro (which gives you timed breaks when your t... |
080cb67a-d054-4abc-b96e-b766af1c4bba | trentmkelly/LessWrong-43k | LessWrong | Swiss Political System: More than You ever Wanted to Know (I.)
Swiss political system may be best known for its extensive use of referenda. However, others may argue that its most striking feature is the ability to avoid political polarization. In this respect it may be unique among the western nations.
That being sa... |
14efb621-2df0-4d38-be6c-e94413ee5bf1 | trentmkelly/LessWrong-43k | LessWrong | Solving adversarial attacks in computer vision as a baby version of general AI alignment
I spent the last few months trying to tackle the problem of adversarial attacks in computer vision from the ground up. The results of this effort are written up in our new paper Ensemble everything everywhere: Multi-scale aggregat... |
fcca63b8-2c8f-4b60-b46e-c5b0abfaee2e | trentmkelly/LessWrong-43k | LessWrong | December 2013 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 th... |
7dfe35ab-8749-4f67-a0ab-e3d52b833f8b | trentmkelly/LessWrong-43k | LessWrong | The ethics of breaking belief
I'm not sure if this is precisely the correct forum for this, but if there is a better place, I don't know what it would be. At any rate...
I'm a student a Catholic university, and there are (as one might surmise) quite a lot of Catholics here, along with assorted other theists (yes, ... |
ce1e29f5-f9f5-4772-96b1-9b3bb884c7c4 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Integrating Artificial Intelligence into Weapon Systems
|
451392b9-d68e-4057-991c-8774199d908a | trentmkelly/LessWrong-43k | LessWrong | Sunscreen. When? Why? Why not?
Hello,
I've been wondering about sunscreen's effectiveness in reducing cancer risk.
The general impression I've gotten in my (admittedly brief) research on it seems: "If you know you'll be exposed for a long duration in the summer/spring, then yes, wear sunscreen. Otherwise, Vitamin D ... |
2be2657d-4464-43c7-8886-dc0b262a9832 | trentmkelly/LessWrong-43k | LessWrong | What are some good Lesswrong-related accounts or hashtags on Mastodon that I should follow?
I used to follow Eliezer Yudkowsky on Twitter. Now I am leaving Twitter and trying Mastodon. I miss Yudkowsky's interesting posts. Whom or what can I follow instead for interesting content? Especially related to overcoming bias... |
8b382d40-0177-45cd-991c-b20b9756a766 | trentmkelly/LessWrong-43k | LessWrong | Infinite Modal Combat: some observations
These notes are from discussions with Kuba Stechly at the Summer Fellows Program, on the topic of infinite modal combat, previously discussed here. In that post, Victoria proved that non-wellfounded infinite modal agents called ProcrastinationBots cooperate with each other.
We... |
d0e199e0-4013-4953-b728-e9d8b976c7cd | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | ELK Proposal - Make the Reporter care about the Predictor’s beliefs
*(This proposal received an honorable mention in the* [*ELK prize results*](https://www.notion.so/ELK-Proposal-Make-the-Reporter-care-about-the-Predictor-s-beliefs-4cb30e79267e4d5591ee7161169170c2)*, and we believe was classified among strategies whic... |
fc730b7d-cf57-45c4-b5f0-f4588d8b4bd9 | trentmkelly/LessWrong-43k | LessWrong | SAEs (usually) Transfer Between Base and Chat Models
This is an interim report sharing preliminary results that we are currently building on. We hope this update will be useful to related research occurring in parallel.
Executive Summary
* We train SAEs on base / chat model pairs and find that SAEs trained on the b... |
e421ae38-913d-4b2a-87bd-fde88d216bb6 | trentmkelly/LessWrong-43k | LessWrong | [link] Essay on AI Safety
I recently wrote an essay about AI risk, targeted at other academics:
Long-Term and Short-Term Challenges to Ensuring the Safety of AI Systems
I think it might be interesting to some of you, so I am sharing it here. I would appreciate any feedback any of you have, especially from others who... |
83cce154-639c-42a2-911a-6b7cbf65550e | trentmkelly/LessWrong-43k | LessWrong | What is wrong with this approach to corrigibility?
Essentially: a button that when you press it kills the AI but instantly gives it reward equal to its expected discounted future reward. And to be clear, this is the AI's estimate of its expected discounted future reward, not some outside estimate.
(This is in the... |
7dacb7b6-df6b-4ff9-9b09-91545a01474a | trentmkelly/LessWrong-43k | LessWrong | Meetup : East Coast Solstice Megameetup
Discussion article for the meetup : East Coast Solstice Megameetup
WHEN: 20 December 2014 03:00:00PM (-0700)
WHERE: 851 Park Place, Brooklyn NY 11216
The weekend of December 20th will be the East Coast Solstice Megameetup. Rationalists and EA folk are invited to visit our gro... |
350351f6-487f-4e43-8225-28a653e8d9fb | trentmkelly/LessWrong-43k | LessWrong | Briefly how I've updated since ChatGPT
I'm laying out my thoughts in order to get people thinking about these points and perhaps correct me. I definitely don't endorse deferring to anything I say, and I would write this differently if I thought people were likely to do so.
1. OpenAI's model of "deploy as early as po... |
b79ba454-e141-4e21-a6c1-f80093bbbb40 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Next week I'm interviewing tech policy expert Teddy Collins who has worked in the White House, DeepMind and CSET. What should I ask him?
Here's Teddy Tantum Collins' [LinkedIn](https://www.linkedin.com/in/tantum-collins-7b916b29), a [recent interview](https://www.youtube.com/watch?v=fkMldjZ-oA0) and [short bio](https:... |
52129718-d456-4cda-9c12-e6c999fb7daa | trentmkelly/LessWrong-43k | LessWrong | What Is Rationalist Berkley’s Community Culture?
Rationalist Status: Wonky (if you are not interested in the rationalist community or mission, this one is not for you)
Epistemic Status: Continuing the discussion
Response to (read this first): The Craft Is Not The Community (Otium)
Sarah’s post, The Craft Is Not The... |
81066094-eae5-49c0-9336-910d2139992f | StampyAI/alignment-research-dataset/lesswrong | LessWrong | The Tree of Life: Stanford AI Alignment Theory of Change
*Update 2023-06-15: Stanford AI Alignment has just wrapped up its first (IMO quite successful) academic year, but it has diverged significantly from some of the plans first laid out here. For updated context, contact me over DM or at* [*gmukobi@cs.stanford.edu*]... |
9c3ea6b8-659e-412b-bf37-fea2e38acb23 | trentmkelly/LessWrong-43k | LessWrong | A developmentally-situated approach to teaching normative behavior to AI
This was submitted to the EthicsNet Guardians' Challenge. I'll be honest here that I hadn't thought much about what EthicsNet is trying to do, but decided to write something and submit to it anyway because it's the sort of approach that seems rea... |
71165954-e975-4908-a503-2f7d8f9bac76 | trentmkelly/LessWrong-43k | LessWrong | Timeless Decision Theory and Meta-Circular Decision Theory
(This started as a reply to Gary Drescher's comment here in which he proposes a Metacircular Decision Theory (MCDT); but it got way too long so I turned it into an article, which also contains some amplifications on TDT which may be of general interest.)
Part... |
abbc0220-dbbe-4303-9f00-796e893e90e5 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Ray Interference: a Source of Plateaus in Deep Reinforcement Learning
1 Introduction
---------------
Deep reinforcement learning (RL) agents have achieved impressive results in recent years, tackling long-standing challenges in board games [[38](#bib.bib38), [39](#bib.bib39)],
video games [[23](#bib.bib23), [26](#b... |
3b899f06-5918-40be-a0cc-12f3207bff47 | trentmkelly/LessWrong-43k | LessWrong | Opinion Article Scoring System
Here I propose a system for scoring media opinion articles. It is part prediction markets - as there is a small amount of money involved - and part forecasting science mechanism design. Journalists that publish an article on the platform must do so with an accompanying stake. Readers (wh... |
d857d06c-ab21-439d-a84e-5732d3655e13 | trentmkelly/LessWrong-43k | LessWrong | Research request (alignment strategy): Deep dive on "making AI solve alignment for us"
We might be able to train AI alignment assistants that massively accelerate/improve the alignment research that gets done. These assistants need not have (strongly) superhuman capabilities or be highly agentic, they just need to be ... |
2f085f79-b824-4ca5-bfca-e5a230ee1cfd | trentmkelly/LessWrong-43k | LessWrong | Internal Target Information for AI Oversight
Thanks to Arun Jose for discussions and feedback.
Summary
In this short post, we discuss the concept of Internal Target Information within agentic AI systems, arguing that agentic systems possess internal information about their targets. This information, we propose, can ... |
b7158298-eece-4dcd-8257-b08c280723be | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | AI Timelines: Where the Arguments, and the "Experts," Stand
*Audio version available at [Cold Takes](https://www.cold-takes.com/where-ai-forecasting-stands-today) (or search Stitcher, Spotify, Google Podcasts, etc. for "Cold Takes Audio")*
[![](https://res.cloudinary.com/cea/image/upload/v1673040427/mirroredImages/7Jx... |
271ab901-9927-4d02-ac82-5563ed341ed8 | trentmkelly/LessWrong-43k | LessWrong | What were the biggest discoveries / innovations in AI and ML?
These could be theoretical breakthroughs (like "the idea of a perceptron" or [something Judea Pearl did]), or they could be watershed developments / new applications that don't necessarily involve much new theory (like AlexNet or AlphaGo). Anything that see... |
b51faf13-465b-4b75-b054-9599bc0afa6d | trentmkelly/LessWrong-43k | LessWrong | Formatting Articles
The comments section has the handy little "help" feature (confusing though its title may be) that explains how to format. I can't figure out where that section is for actual articles, which use completely different syntax.
In particular, I don't know how to post a I'm having trouble posting "lyric... |
500a82b5-dd7c-4f3e-b2fe-eae4ecdc9ce3 | trentmkelly/LessWrong-43k | LessWrong | Who observes being you, and do they cheat in SIA?
Warning: this post is somewhat technical – looking at this summary should help.
> 1,000,000 people are in a giant urn. Each person is labeled with a number (number 1 through number 1,000,000).
> A coin will be flipped. If heads, Large World wins and 999,999 people wil... |
0139d1c8-ae92-4a64-bee0-6f890d9d37d3 | trentmkelly/LessWrong-43k | LessWrong | Is it possible to prevent the torture of ems?
When I was reading The Seven Biggest Dick Moves in the History of Gaming, I was struck by the number of people who are strongly motivated to cause misery to others [1], apparently for its own sake. I think the default assumption here is that the primary risk to ems is from... |
455ad421-1a8d-4df9-b5e1-2b2b53fcf66c | trentmkelly/LessWrong-43k | LessWrong | Animated explainer video promoting EA-themed effective giving ideas and meta-charities
|
d32e9188-acd3-44a3-b9ed-60d8cb72854d | trentmkelly/LessWrong-43k | LessWrong | You can validly be seen and validated by a chatbot
There’s a common sentiment saying that a chatbot can’t really make you feel seen or validated. As chatbots are (presumably) not sentient, they can’t see you and thus can’t make you seen either. Or if they do, it is somehow fake and it’s bad that you feel that way.
So... |
5dbf93db-fb2f-43e0-b473-4ccf118ab4e4 | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | Using AI to accelerate scientific discovery - Demis Hassabis (Crick Insight Lecture Series)
hello
well welcome everybody it's almost
normal
almost but when i look around the room
and see all these masks isn't quite
normal but this is the start of
normality so let's celebrate it and
we've got a great start with demis
h... |
1c71a662-4d42-4161-a9c3-6a37086e48bf | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | How-to Transformer Mechanistic Interpretability—in 50 lines of code or less!
*Produced as part of the*[*SERI ML Alignment Theory Scholars Program*](https://serimats.org/) *- Winter 2022 Cohort.*
What if I told you that in just one weekend you can get up to speed doing practical Mechanistic Interpretability research o... |
ea2d1635-7559-4e50-ad2b-705391e40a9d | trentmkelly/LessWrong-43k | LessWrong | Is Stupidity Expanding? Some Hypotheses.
To be explained: It feels to me that in recent years, people have gotten stupider, or that stupid has gotten bigger, or that the parts of people that were always stupid have gotten louder, or something like that.
I’ve come up with a suite of hypotheses to explain this (with a ... |
6b9a5d7b-4bf5-4a5b-88a4-e1e8e4ecd63d | trentmkelly/LessWrong-43k | LessWrong | Is frequentism compatible with the principle of indifference in any meaningful sense?
None |
4e6df41c-5e13-4ff5-8f72-52e7a1fb3cff | trentmkelly/LessWrong-43k | LessWrong | My Best Case vs Your Worst Case
Is there a name for the (I claim) extremely common practice of blithely and unconsciously always looking at your own view (political especially) in terms of its best possible outcomes, while always characterizing an opposing point of view by its worst possibilities?
If not, I think the... |
32dc151d-b692-4508-a39f-b66042736f03 | trentmkelly/LessWrong-43k | LessWrong | Covid 8/13: Same As It Ever Was
Last time: Covid 8/6: The Case of the Missing Data
I got the opportunity during this past week to work with a state official to try and solve the case of the missing data. It felt great to have the help and also to potentially be reaching someone who could make a difference, even if th... |
69c78b1f-115e-4275-81bc-eb5f084abbd8 | trentmkelly/LessWrong-43k | LessWrong | One week left to apply for the Roots of Progress Blog-Building Intensive
The application deadline for the the 2024 cohort of The Roots of Progress Blog-Building Intensive is Friday, June 7—just over a week away. If you want to apply, do it now.
The Blog-Building Intensive is an eight-week program for aspiring progres... |
f62994ef-ab11-4fef-bd5e-f56362022bb6 | trentmkelly/LessWrong-43k | LessWrong | Product orientation
When I applied to graduate schools, I methodically went through over 50 bioinformatics program websites and US News and World Report to find their tuition data and ranking, then threw it all into a spreadsheet. I included cost of living, the length of the program, and ran a linear regression to fig... |
c9942dbe-60e6-4247-9b15-442efc07accf | trentmkelly/LessWrong-43k | LessWrong | Useful Vices for Wicked Problems
I've claimed that the best way to learn is by writing about important topics. (Examples I've worked on include: which charity to donate to, whether life has gotten better over time, whether civilization is declining, whether AI could make this the most important century of all time for... |
9c79ff5b-017e-41c7-b177-0440996f80a5 | trentmkelly/LessWrong-43k | LessWrong | Self Improvement - All encompassing vs. Focused
p.p1 {margin: 0.0px 0.0px 0.0px 0.0px; font: 12.0px Helvetica} p.p2 {margin: 0.0px 0.0px 0.0px 0.0px; font: 12.0px Helvetica; min-height: 14.0px}
Lately I've been identifying a lot of things about myself that need improvement and thinking about ways to fix them. This po... |
9471b997-de9d-4f56-b7e6-ca58b8c77539 | trentmkelly/LessWrong-43k | LessWrong | Recognizing Optimality
Short feedback loops are great when the feedback is clear, but without a guiding framework it can be hard to know what to make of the feedback -- especially in the short term. If things get uncomfortable, is that a bad thing or is that a counter-steering? If I'm getting results quickly, am I Goo... |
f792776d-1aa9-457b-b586-74eaa35c79a2 | trentmkelly/LessWrong-43k | LessWrong | Robustness to fundamental uncertainty in AGI alignment
Abstract: The AGI alignment problem has a bimodal distribution of outcomes with most outcomes clustering around the poles of total success and existential, catastrophic failure. Consequently, attempts to solve AGI alignment should, all else equal, prefer false neg... |
9a85e1c8-311c-4c31-ba7a-5462055f8047 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | AXRP Episode 18 - Concept Extrapolation with Stuart Armstrong
[Google Podcasts link](https://podcasts.google.com/feed/aHR0cHM6Ly9heHJwb2RjYXN0LmxpYnN5bi5jb20vcnNz/episode/YmE3NDc2ZmYtMmZlMy00YmEzLThjODQtNTBiYzYwMjQ0NGVj)
Concept extrapolation is the idea of taking concepts an AI has about the world - say, “mass” or ... |
9a5e0b67-9d91-45b2-bd68-a14d0a44b624 | trentmkelly/LessWrong-43k | LessWrong | Systematic Sandbagging Evaluations on Claude 3.5 Sonnet
This was the project I worked on during BlueDot Impact's AI Safety Fundamentals Alignment course, which expands on findings from Meinke et al's "Frontier Models are Capable of In-context Scheming".
Summary
* A dataset of 1,011 variations of the sandbagging pro... |
15786130-f749-48d8-bb80-5d02beae0ac1 | trentmkelly/LessWrong-43k | LessWrong | Interest in video-conference discussion about sequences and/or virtual meetups?
Update: We're up to 9 participants. Per jsalvatier's suggestion, I have created a Google Group for this. It looks like we have 8-9 total, and the move toward finalizing groups and dates/times is now at the Google Group thread HERE.
I plan... |
dd39561d-49b3-41b1-a8b1-6646a31c5232 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "Last Friday, I finally packaged up the quarterly release of JCVI's automatic prokaryote functional annotation pipeline and distributed it to the other 3 sequencing centers for the Human Microbiome Project. It looks at genes found in newly-sequenced bacterial genomes and guesses what they do. As often happens when I re... |
112bc3d1-5931-44c9-8542-e80bd828784d | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post953
Thanks to Michael Ripa for feedback. TL;DR There are two types of problems that AI systems can have: problems that we can observe during development and problems that we can’t. The hardest part of the alignment problem involves problems we can’t observe. I outline seven types of solutions for unobservable ... |
b30e54e9-7a42-4f5c-a9e6-5699317f2698 | trentmkelly/LessWrong-43k | LessWrong | One possible approach to develop the best possible general learning algorithm
One possible approach to develop the best possible general learning algorithm
Author: Martí Llopart
Introduction
In the world we live in, there are many problems that need to be urgently solved: climate change[1], the risk of a new pandem... |
8d331ed6-c607-41aa-a5bf-4da20f7bffd2 | trentmkelly/LessWrong-43k | LessWrong | How is the world different if I make a donation?
How is the world different if I give X dollars to the Against Malaria Foundation, where X < 100,000 USD? (In reality, giving directly to GiveWell is probably the better choice. But using AMF works well for this question.)
Here's the answer that would make me very motiv... |
dfb88f48-f113-4185-aa85-2f5f70c8633c | trentmkelly/LessWrong-43k | LessWrong | Ben Goertzel's "Kinds of Minds"
I'd like to recommend Ben Goertzel's book from 2006 Hidden Pattern, and particularly Chapter 2 "Kinds of Minds". (The book is also available at the author's site, but that may trigger SSL certificate warnings.)
We've come to understand that a future AGI may have non-anthropomorphic im... |
e8b34a50-d9ba-42b9-b4fa-176a83b498f6 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Against GDP as a metric for timelines and takeoff speeds
Or: Why AI Takeover Might Happen Before GDP Accelerates, and Other Thoughts On What Matters for Timelines and Takeoff Speeds
============================================================================================================================
*[*[*Crossp... |
2eaed9b9-b80b-40ec-9dd5-f062f18409d2 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post861
Produced as part of the SERI ML Alignment Theory Scholars Program - Summer 2023 Cohort, under the mentorship of Jeffrey Ladish. I'm grateful to Palisade Research for their support throughout this project. tl;dr: demonstrating that we can cheaply undo safety finetuning from open-source models to remove refu... |
aa4e517c-722d-45cd-ad97-19b309356727 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | The First World Takeover
Before Robin and I move on to talking about the Future, it seems to me wise to check if we have disagreements in our view of the Past. Which might be much easier to discuss - and maybe even resolve... So...
In the beginning was the Bang. For nine billion years afterward, nothing much happe... |
31c7430e-b130-4e9e-b6f4-704482564395 | StampyAI/alignment-research-dataset/arbital | Arbital | Conditional probability: Refresher
$\mathbb P(\text{left} \mid \text{right})$ is the probability that the thing on the left is true _assuming_ the thing on the right is true, and it's defined as $\frac{\mathbb P(\text{left} \land \text{right})}{\mathbb P(\text{right})}.$
Thus, $\mathbb P(yellow \mid banana)$ is the p... |
e34906ea-0d47-4b9d-92b0-4e3e98027056 | trentmkelly/LessWrong-43k | LessWrong | Paper review: “The Unreasonable Effectiveness of Easy Training Data for Hard Tasks”
TL;DR: Scalable oversight seems easier based on experiments outlined in a recent paper; questions arise about the implications of these findings.
The following graciously provided feedback and advice on the draft, for which I am d... |
0a58c534-5367-4ad0-9ad8-2ea50439395b | StampyAI/alignment-research-dataset/blogs | Blogs | Importance motivation: a double-edged sword
Working on something important is often stressful. A hypothesis came up in conversation (h/t Michael Vassar) that labeling your work as important actually decreases the quality of the output. I have seen this effect in action quite a bit, e.g. when someone can write an essay... |
195d2e02-748f-448c-92b9-0ccff237ba64 | trentmkelly/LessWrong-43k | LessWrong | Open Mic - August 2023
In stand up comedy, an open mic is a place that is open to amateurs. A place to experiment with material. A place to babble.
Currently, on LessWrong, personal blog posts, open threads, and shortforms are all places where you are invited to babble. However, I get the impression that there is a c... |
1ca409fa-f2fe-4247-b7b6-740be32e9bed | StampyAI/alignment-research-dataset/blogs | Blogs | Photos from the first AI Safety Camp
[[See image gallery at aisafety.camp](https://aisafety.camp/2018/12/08/aisc1_photos/)] |
c70e1fb4-fd33-4dc4-9777-7b021ce9602b | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | [Weekly Event] Alignment Researcher Coffee Time (in Walled Garden)
I'm organizing a weekly one hour coffee time for alignment researchers to talk about their research and what they're interested about, every Monday starting tomorrow. The time (9pm CEST, see [here](https://everytimezone.com/s/0d39d311) for your timezon... |
02626d4c-9837-4c19-8cac-ad9b9ccbf13c | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | [MLSN #6]: Transparency survey, provable robustness, ML models that predict the future
*You can* [*subscribe to the newsletter here*](https://newsletter.mlsafety.org/)*, follow the newsletter on* [*twitter*](https://twitter.com/ml_safety) *here, or join the subreddit* [*here*](https://www.reddit.com/r/mlsafety/)*.*
... |
2d8614bf-5793-493e-a00a-521c678753c5 | StampyAI/alignment-research-dataset/distill | Distill Scientific Journal | A Discussion of 'Adversarial Examples Are Not Bugs, They Are Features': Two Examples of Useful, Non-Robust Features
#rebuttal,
.comment-info {
background-color: hsl(54, 78%, 96%);
border-left: solid hsl(54, 33%, 67%) 1px;
padding: 1em;
color: hsla(0, 0%, 0%, 0.67);
}
#header-info {
margin-top: 0;
margin-bott... |
27831c1c-8e57-4842-9bdf-b52a8a34737d | trentmkelly/LessWrong-43k | LessWrong | Harry Potter and the Methods of Rationality discussion thread, part 24, chapter 95
This is a new thread to discuss Eliezer Yudkowsky’s Harry Potter and the Methods of Rationality and anything related to it. This thread is intended for discussing chapter 95. The previous thread has passed 300 comments.
There is now a... |
e445d788-6687-4170-b971-81d67f4340d9 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Tutor-GPT & Pedagogical Reasoning
*This post hopes foster discussion about latent heutagogic potential in human+LLM collaboration and share some early results with the LessWrong community.*
*Spontaneous emergence of theory of mind capabilities in large language models hints at a host of overhung possibilities relevan... |
5815ec79-1dc8-49a5-9a34-4a7bf4e794c5 | trentmkelly/LessWrong-43k | LessWrong | Using lying to detect human values
|
134b3d8c-fad8-4003-b8ef-2635a56c666f | StampyAI/alignment-research-dataset/blogs | Blogs | Trends in the cost of computing
*Posted 10 Mar 2015*
Computing power available per dollar has probably increased by a factor of ten roughly every four years over the last quarter of a century (measured in FLOPS or MIPS).
Over the past 6-8 years, the rate has been slower: around an order of magnitude every 10-16 ye... |
201a9975-c7f5-40d9-8697-42462b796180 | trentmkelly/LessWrong-43k | LessWrong | A survey of the top posters on lesswrong
In a post recently someone mentioned that there was a list of "Top 15" posters by karma. That inspired me to send all of them this note:
> Hi,
>
> I am messaging you (now) because you are one of the 15 top contributors of the past 30 days of LW.
>
> I was wondering if you d... |
30ba65fb-a624-46c9-bee4-f1405736099e | trentmkelly/LessWrong-43k | LessWrong | On what rationality-related topic should I give a school presentation?
My school has a weekly event on Thursdays where someone can give a 15-25 minute lecture about a topic of their choice during the lunch break. The standard attendance is about 20-30, aged between 14 and 18, and some teachers drop by if the topic is ... |
82097348-c1b0-4815-9277-76a820e760bc | trentmkelly/LessWrong-43k | LessWrong | Are LLMs on the Path to AGI?
I am unsure, but I disagree with one argument that they aren’t.
There’s a joke about how humans have gotten so good at thinking that they tricked rocks into thinking for them. But it’s a joke, in part because it’s funny to say that computers work by “tricking rocks into thinking,” and in p... |
170e200c-68f0-483d-8b94-72b0a0a1a109 | trentmkelly/LessWrong-43k | LessWrong | Strong moral realism, meta-ethics and pseudo-questions.
On Wei_Dai's complexity of values post, Toby Ord writes:
> There are a lot of posts here that presuppose some combination of moral anti-realism and value complexity. These views go together well: if value is not fundamental, but dependent on characteristics of ... |
c6a2d835-248b-4cbd-a2e2-63d42b5255a6 | trentmkelly/LessWrong-43k | LessWrong | Inconsistent math is great
Consistency is nice to have as a guarantee, but usually you don't need something so strong. A lot of weird things in math just collapse if the underlying object is finite.
Graham Priest gets into this a lot. |
73d9a009-b7a4-42a7-a526-24820bb859fb | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Bayesian Persuasion?
I'm an economist and quite new to AI alignment. In reading about the perils of persuasive AI, I was reminded of an influential model in economic theory: the [Bayesian persuasion model](https://web.stanford.edu/~gentzkow/research/BayesianPersuasion.pdf) (Kamenica and Gentzkow, 2011). It's used to m... |
ae5a7ce2-89a6-4061-a611-8b4a45d416b1 | trentmkelly/LessWrong-43k | LessWrong | Is there an equivalent of the CDF for grading predictions?
When I see people grading their predictions, it's always by: (a) bucketing their predictions by probability (into a "46-55%" bucket, a "56-75%" bucket, ...), and then (b) plotting each bucket's nominal probability vs empirical frequency-of-correctness. See e.g... |
22a3d6dc-1b6a-4546-b075-65e302296de6 | trentmkelly/LessWrong-43k | LessWrong | Sunzi's《Methods of War》- Planning Attacks
This is a translation of Chapter 3 of The Art of War by Sunzi. No English sources were used.
----------------------------------------
> 孙子曰：夫用兵之法，
>
> 全国为上，破国次之，
>
> 全军为上，破军次之；
>
> 全旅为上，破旅次之；
>
> 全卒为上，破卒次之；
>
> 全伍为上，破伍次之。
* It is best to conquer a state whole rather t... |
bf404aaf-12ee-48e5-92a3-2d12f8fe8fa3 | trentmkelly/LessWrong-43k | LessWrong | [SP] The Edge of Morality
Scott Alexander writes that philosophy is the art of exploring the edge cases of our ethics. It is clear to most that one should not kill innocent people for pleasure, but that will not help us uncover new insights. Instead, we look at the edge cases where it is not clear what morality says. ... |
2259d212-9c07-4d3a-9df1-c0e547537936 | trentmkelly/LessWrong-43k | LessWrong | February 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 th... |
5bedfdef-59e9-40cb-8f46-f5a589b477c6 | trentmkelly/LessWrong-43k | LessWrong | Disability Culture Meets the Transhumanist Condition
With apologies to Ed Regis.
Modern science has caused humankind to develop better cures and patches for once-debilitating conditions; people often survive maladies which would have killed them not long ago. In the wake of this and of a recently changing attitude re... |
bce7936d-38f6-4388-9f0e-db0b1959625f | trentmkelly/LessWrong-43k | LessWrong | SB 1047 gets vetoed
Just what it says on the tin. Covered most everywhere, but I found the quote in this Reuters article stuck out to me the most:
> Newsom said the bill "does not take into account whether an AI system is deployed in high-risk environments, involves critical decision-making or the use of sensitive da... |
79effa2f-592d-42ed-a32e-a0a2f8cdf284 | trentmkelly/LessWrong-43k | LessWrong | EA & LW Forums Weekly Summary (5 - 11 Sep 22')
Supported by Rethink Priorities
Cross-posted on the EA forum here.
This is part of a weekly series - you can see the full collection here. The first post includes some details on purpose and methodology.
If you'd like to receive these summaries via email, you can subsc... |
327414cd-fa1d-40ad-b253-c664394876e0 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | [AN #146]: Plausible stories of how we might fail to avert an existential catastrophe
Alignment Newsletter is a weekly publication with recent content relevant to AI alignment around the world.
HIGHLIGHTS
==========
**[What Multipolar Failure Looks Like, and Robust Agent-Agnostic Processes (RAAPs)](https://www.align... |
6f3ff8ab-77ef-4884-bb30-cf20350ec165 | trentmkelly/LessWrong-43k | LessWrong | CHCAI/MIRI research internship in AI safety
The Center for Human-Compatible AI (CHCAI) and the Machine Intelligence Research Institute (MIRI) are looking for talented, driven, and ambitious technical researchers for a summer research internship.
Background:
CHCAI is a research center based at UC Berkeley with PIs... |
2ff3e061-f639-407e-bca1-9bcbc5f320b4 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Accumulating Risk Capital Through Investing in Cooperation
1. Introduction
----------------
Sequential social dilemmas (SSDs) are games where short-term individual incentives conflict with the long-term social good. Such games are pervasive and describe situations in which we would deploy automated multi-agent syst... |
0b1429c3-574f-44b1-9222-f5a43a711176 | trentmkelly/LessWrong-43k | LessWrong | The Uncertainty of Death and Taxes
Follow-up of sorts to: Easy Mode and Hard Mode, also Spoiler-Free Review: Death and Taxes
Epistemic Status: Early exploration of concepts that deserve a better, tighter treatment at a future time. This has been in my drafts for a while, figure I’ll push it out there and then redo th... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.