id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
7db823d8-435d-4fd3-8ac1-3a33af0c70ad | trentmkelly/LessWrong-43k | LessWrong | In Defence of Simple Ideas That Explain Everything But Are Wrong
I've been thinking, and writing, about The Impossible Question of the Thyroid for some while now.
I came up with what I thought was a good stab at an answer to its majestic mystery:
http://lesswrong.com/r/discussion/lw/nef/the_thyroid_madness_core_argu... |
029118df-e94a-4693-9625-0a3c2f72056f | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Mesa-Optimization: Explain it like I'm 10 Edition
There are a [couple](https://www.lesswrong.com/posts/XWPJfgBymBbL3jdFd/an-58-mesa-optimization-what-it-is-and-why-we-should-care) [of](https://ui.stampy.ai?state=8160_) [explanations](https://www.lesswrong.com/tag/mesa-optimization) of mesa-optimization available. I th... |
b921609f-7d1f-4ecf-ae25-1dd4551ff323 | trentmkelly/LessWrong-43k | LessWrong | How is rationalism different from utilitarianism?
Would appreciate either theoretical explanations or illustration using examples. I am new to the forum, and it appears to me that both rationalists and utilitarianists would use "utility maximization" as the key decision criterion. Maybe for rationalists, "utility" is ... |
e6f86f6a-8ae4-4a85-8bd0-c2a4023e6bb0 | trentmkelly/LessWrong-43k | LessWrong | Can coherent extrapolated volition be estimated with Inverse Reinforcement Learning?
Given the following conditions, is it possible to approximate the coherent extrapolated value of humanity to a "good enough" level?:
* Some form of reward/cost function estimation is used, such as inverse reinforcement learning or i... |
92a030be-9d1e-4c0c-aecd-c219a5f26802 | StampyAI/alignment-research-dataset/special_docs | Other | Choice Set Misspecification in Reward Inference
Choice Set Misspecification in Reward Inference
Rachel Freedman,Rohin Shah and Anca Dragan
University of California, Berkeley
frachel.freedman, rohinmshah, anca g@berkeley.edu,
Abstract
Specifying reward functions for robots that operate
in environments without a natural ... |
38a59100-b2e8-44f4-99c7-d7005458ee77 | trentmkelly/LessWrong-43k | LessWrong | Sentinel's Global Risks Weekly Roundup #12/2025: Famine in Gaza, H7N9 outbreak, US geopolitical leadership weakening.
Executive summary
Forecasters believe there’s an 18% chance (range: 4%-50%) that there will be a famine in any part of Gaza by the end of 2025, according to the UN and its Integrated Food Security Phas... |
587a0134-3d89-45cc-bff8-905e2100463a | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Language models seem to be much better than humans at next-token prediction
[Thanks to a variety of people for comments and assistance (especially Paul Christiano, Nostalgebraist, and Rafe Kennedy), and to various people for playing the game. Buck wrote the top-1 prediction web app; Fabien wrote the code for the perpl... |
2b101fd0-24a9-4570-be4a-9c9b2c2ddd35 | trentmkelly/LessWrong-43k | LessWrong | Cultivating And Destroying Agency
Something that really surprised me about high school is how little agency the majority of the students there had. This was even more pronounced among the students at the local magnet school, Academies of Loudoun; they were selected for on intelligence and work ethic, but generally had... |
f42e97e4-08fd-4a65-b2db-e19e3e1c065f | trentmkelly/LessWrong-43k | LessWrong | Is it a good idea to use Soylent once/twice a day?
Edit: Sorry, I didn't realize there has been so much discussion on this already! I thought I had just stumbled across some obscure product haha. Anyway, I've been reading through discussions here, on Hacker News, Tim Ferris' blog etc. There's been a lot of talk about ... |
6991eaac-8af3-481e-9374-67135dd6701c | trentmkelly/LessWrong-43k | LessWrong | Julia Tasks 101
Tasks are how Julia handles parallelism & concurrency. Tasks are defined at the program level and Julia's scheduler maps them to hardware/OS threads.
Tasks have many names in other languages: "symmetric coroutines, lightweight threads, cooperative multitasking, or one-shot continuations". They're part... |
80e561eb-e7ca-481c-b0ca-2b37884cf9d3 | trentmkelly/LessWrong-43k | LessWrong | Apply to Aether - Independent LLM Agent Safety Research Group
The basic idea
Aether will be a small group of talented early-career AI safety researchers with a shared research vision who work full-time with mentorship on their best effort at making AI go well. That research vision will broadly revolve around the align... |
50f3bc43-5e49-4068-ad35-bb059df85cc6 | trentmkelly/LessWrong-43k | LessWrong | Alpha Go Zero comments
DeepMind has published a new paper in Nature detailing "Alpha Go Zero", a Go AI that was trained with only self-play. Alpha Go Zero equalled the ability of a system trained with supervised learning on a Go profession game corpus in 24 hours (using the same computing resources) and exceeded the a... |
a3557998-ef42-41d2-8af5-fe95ac7a132a | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Another problem with AI confinement: ordinary CPUs can work as radio transmitters
Excerpts:
> Computer tech expert William Entriken has created a C library that can make a computer emit radio waves even if the device doesn't include any radio transmission hardware.
>
> His program, called System Bus Radio, is writt... |
ddea0b31-5c26-4022-b3fc-6b1a13d786d3 | trentmkelly/LessWrong-43k | LessWrong | Weekly LW Meetups
This summary was posted to LW Main on January 20th. The following week's summary is here.
Irregularly scheduled Less Wrong meetups are taking place in:
* Sao Paulo - Meetup de janeiro: 28 January 2017 02:00PM
The following meetups take place in cities with regular scheduling, but involve a change... |
79fef311-5ddb-4604-ba8c-c59db6e64f1a | trentmkelly/LessWrong-43k | LessWrong | Contra: Avoiding Sore Arms
A common issue among when dancing Lark role is sore right arms/wrists after dancing. Why does this happen, and what can we do about it?
During the swing, the dancers need to be held together or they'll fall over backwards. That's not because they're leaning back (stay upright!) but because ... |
9e726277-74e6-4adb-a315-f5994945612d | trentmkelly/LessWrong-43k | LessWrong | Meditation and Neuroscience, some odds and ends
[cross-posted to my blog]
[About me: I'm the meditationstuff guy and the "folding" guy. I have a PhD in bioengineering, during which I did work with human clinical EEG (and also audited neuroscience and machine learning classes), but I'm not a neuroscientist, and I didn... |
fdf40639-4170-4d66-8dfd-88b43f93f5de | trentmkelly/LessWrong-43k | LessWrong | Neural networks generalize because of this one weird trick
> Produced under the mentorship of Evan Hubinger as part of the SERI ML Alignment Theory Scholars Program - Winter 2022 Cohort
> A big thank you to all of the people who gave me feedback on this post: Edmund Lau, Dan Murfet, Alexander Gietelink Oldenziel, Luc... |
668a247e-e34c-4d16-a3df-9e6ded8c9db3 | trentmkelly/LessWrong-43k | LessWrong | OB Meetup: Millbrae, Thu 21 Feb, 7pm
The Overcoming Bias meetup has been scheduled for Thursday, February 21st, at 7pm. We're going to look at locating this in Millbrae within walking distance of the BART / Caltrain station. The particular restaurant I had in mind turns out to be booked for Thursdays, so if you know... |
2ecf0694-4f64-474a-9b2b-36dd49b650d7 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Asymptotically Unambitious AGI
Original Post:
We [present an algorithm](http://www.tinyurl.com/bomai-anon) [[updated version](https://arxiv.org/abs/1905.12186)], then show (given four assumptions) that in the limit, it is human-level intelligent and benign.
Will MacAskill has commented that in the seminar room, he i... |
aad61846-8e12-422f-9e7c-a5db015f75fe | trentmkelly/LessWrong-43k | LessWrong | Have epistemic conditions always been this bad?
In the last few months, I've gotten increasingly alarmed by leftist politics in the US, and the epistemic conditions that it operates under and is imposing wherever it gains power. (Quite possibly the conditions are just as dire on the right, but they are not as visible ... |
1e6ce6ad-84d0-4434-bcd2-492c20ee00eb | trentmkelly/LessWrong-43k | LessWrong | Why wasn't preservation with the goal of potential future revival started earlier in history?
Cross-posted from my blog, Neurobiology Notes.
John Hunter (1728–1793) did not have an especially promising start to his academic life. He was born the youngest of 10 children to a family living in the countryside near Glas... |
f50a3d2f-0240-41bd-835a-6d178b90e6bb | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Counterfactuals, thick and thin
*Summary: There's a "thin" concept of counterfactual that's easy to formalize and a "thick" concept that's harder to formalize.*
Suppose you're trying to guess the outcome of a coinflip. You guess heads, and the coin lands tails. Now you can ask how the coin *would* have landed if you ... |
91ebac91-ee9e-4498-a2e9-1bb715ef7642 | trentmkelly/LessWrong-43k | LessWrong | Imbue (Generally Intelligent) continue to make progress
I've been following the company Imbue and their podcast Generally Intelligent since they started. They've said thoughtful and creative things in their podcast, and I think they are making impressive progress towards AGI considering their relatively smaller size.
... |
aa04be5b-9a5e-44f6-a9e2-2247a87c2e9d | trentmkelly/LessWrong-43k | LessWrong | When Does Altruism Strengthen Altruism?
Joey Savoie recently wrote that Altruism Sharpens Altruism:
> I think many EAs have a unique view about how one altruistic action affects the next altruistic action, something like altruism is powerful in terms of its impact, and altruistic acts take time/energy/willpower; th... |
fba315a7-d4cb-4d4d-b22b-d13c97a083f9 | trentmkelly/LessWrong-43k | LessWrong | An anthropomorphic AI dilemma
[Metadata: crossposted from https://tsvibt.blogspot.com/2023/01/the-anthropomorphic-ai-dilemma.html. First completed January 21, 2023.]
Either generally-human-level AI will work internally like humans work internally, or not. If generally-human-level AI works like humans, then takeoff ca... |
df9866ab-30a9-4353-b1a7-cd8023e64f4a | trentmkelly/LessWrong-43k | LessWrong | Nick Bostrom's TED talk and setting priorities
I just watched Nick Bostrom's TED talk titled "Humanity's biggest problems aren't what you think they are." I was expecting a talk giving Bostrom's take on what he thought the three biggest existential (or at least catastrophic) risks are, but instead, "existential risk" ... |
6d4c435a-238d-4954-8704-5294f21fe5e6 | trentmkelly/LessWrong-43k | LessWrong | LessWrong meetup, London UK, 2010-06-06 16:00
The next London LessWrong meetup will be at 16:00 on Sunday 6 June in the Shakespeare's Head near Holborn station. I'll put a Less Wrong sign on the table so you can find us; I look like this. Send me a direct message with your mobile number or email me (paul at ciphergo... |
54f94da5-8ef4-444f-aee0-0418fb482145 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Pretraining Graph Neural Networks for few-shot Analog Circuit Modeling and Design.
I Introduction
---------------
The
semiconductor industry largely owes its rapid progress in the last century to the scaling of silicon process technology, which has consistently allowed designers to fit more processing power into th... |
ac65501b-9a52-4631-afd4-6836092f3e28 | StampyAI/alignment-research-dataset/blogs | Blogs | Advise MIRI with Your Domain-Specific Expertise
MIRI currently has a few dozen volunteer advisors on a wide range of subjects, but we need more! If you’d like to help MIRI pursue its mission more efficiently, please [sign up to be a MIRI advisor](https://docs.google.com/spreadsheet/viewform?formkey=dG1oUmNybktpUzBXY0J... |
93fe6cf5-36d6-4660-8678-417413bca6b3 | trentmkelly/LessWrong-43k | LessWrong | Idealized Factored Cognition
(This post is part of a sequence that's meant to be read in order; see the preface.)
1. HCH and Ideal Debate
Recall from post #-2 that we have two perspectives on stock IDA.[1] One is that of a human with access to a model, the other is that of an HCH tree.[2] We can think of HCH as 'pur... |
12c2fcf6-46c4-4a06-8ee9-c4fac6aec7f5 | trentmkelly/LessWrong-43k | LessWrong | Rational Ethics
[Looking for feedback, particularly on links to related posts; I'd like to finish this out as a post on the main, provided there aren't too many wrinkles for it to be salvaged.]
Morality as Fixed Computation, Abstracted Idealized Dynamics, as part of the Metaethics Sequence, discuss ethics as computat... |
f1b8178d-54a3-4633-9e3f-c779700a9c2b | trentmkelly/LessWrong-43k | LessWrong | February 27 2011 Southern California Meetup
There will be a LessWrong meetup for Southern California on Sunday February 27th at 1:30PM. The location is being kindly provided by the Harvey Mudd A.I. & Future Technology Club in the Platt Campus Center of Harvey Mudd College. There will food, drinks, small group conver... |
b40ebdad-5af4-4c1b-b4fa-0bb157cc2a52 | trentmkelly/LessWrong-43k | LessWrong | Steering Llama-2 with contrastive activation additions
The effects of subtracting or adding a "sycophancy vector" to one bias term.
TL;DR: By just adding e.g. a "sycophancy vector" to one bias term, we outperform supervised finetuning and few-shot prompting at steering completions to be more or less sycophantic. Furt... |
dc3e991e-f62d-4926-8b17-4ea7225eb192 | trentmkelly/LessWrong-43k | LessWrong | Old Airports
One thing I enjoy about aerial imagery is that you can see traces of how cities used to be. Tracing the spaces left behind by old train lines is a lot of fun, but recently I've been looking at some airports:
Old Saint-Pierre Airport (link), now with housing:
Old Jijiga Airport (link), also with housing,... |
46caf75e-86bc-439a-ae16-660adb9898d0 | trentmkelly/LessWrong-43k | LessWrong | Moral public goods
Automatically crossposted
Suppose that a kingdom contains a million peasants and a thousand nobles, and:
* Each noble makes as much as 10,000 peasants put together, such that collectively the nobles get 90% of the income.
* Each noble cares about as much about themselves as they do about all pea... |
5bb34e9c-9875-4afd-b057-f602638ed523 | trentmkelly/LessWrong-43k | LessWrong | Chapter 37: Interlude: Crossing the Boundary
It was almost midnight.
Staying up late was simple enough for Harry. He just hadn't used the Time-Turner. Harry followed a tradition of timing his sleep cycle to make sure he was awake for when Christmas Eve turned into Christmas Day; because while he'd never been young en... |
b9bfc0bf-b4c8-4049-9fa3-f1664cee9ddb | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Shard Therapy, preambe
In the [last post](https://www.lesswrong.com/posts/MMygRZyMJweamcrJc/the-human-decision-making-algorithm-s) we talked about natural selection's two main ways of approximating utility functions in humans: Fixed stimulus reactions, and reinforcement learning. In this post I want to hone in on rein... |
bdb2b2ea-8b22-40bc-9d6e-6b60127b062d | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Longtermist reasons to work for innovative governments
*Epistemic status: speculative*
Longtermists have good reasons to be interested in institutional innovation. [Broad longtermist institutions](https://forum.effectivealtruism.org/posts/op93xvHkJ5KvCrKaj/institutions-for-future-generations) could focus on improv... |
74f46059-8f52-4475-a043-f44ddcc20357 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | If I were a well-intentioned AI... I: Image classifier
Introduction: If I were a well-intentioned AI...
================================================
I've often [warned](https://intelligence.org/smarter-than-us/) people about the dangers of [anthropomorphising](https://www.lesswrong.com/posts/f4RJtHBPvDRJcCTva/wh... |
86588ddd-20f6-4af1-9beb-4f6518a370fd | trentmkelly/LessWrong-43k | LessWrong | One model of understanding independent differences in sensory perception
This week my friend Anna said to me; "I just discovered my typical mind fallacy around visualisation is wrong". Naturally I was perplexed and confused. She said;
“When I was in second grade the teacher had the class do an exercise in visuali... |
996e068f-fde5-42f4-bacf-e32bc41bbfe7 | trentmkelly/LessWrong-43k | LessWrong | Defining Myopia
IID vs Myopia
In a comment to Partial Agency, Rohin summarized his understanding of the post. He used the iid assumption as a critical part of his story. Initially, I thought that this was a good description of what was going on; but I soon realized that iid isn't myopia at all (and commented as such).... |
eb15a317-f82d-4c29-bacf-33092360121b | trentmkelly/LessWrong-43k | LessWrong | A Poorly Planned Loft Bed
Planning Lily's loft bed reminded me of a much more poorly constructed one I built in 2004, around when I started writing here. I had just started college and was sorting out a room layout with my roommate. The beds were three pieces: two vertical "H" pieces, and a metal frame to connect them... |
3049b8a8-5b21-48bb-89e7-f6c00fdc2a78 | StampyAI/alignment-research-dataset/blogs | Blogs | Historic trends in structure heights
Trends for tallest ever structure heights, tallest ever freestanding structure heights, tallest existing freestanding structure heights, and tallest ever building heights have each seen 5-8 discontinuities of more than ten years. These are:
* **Djoser and Meidum pyramids** (~2600... |
52609a45-969b-4482-afaf-b84b2c7e4daa | trentmkelly/LessWrong-43k | LessWrong | The weakest arguments for and against human level AI
While going through the list of arguments for why to expect human level AI to happen or be impossible I was stuck by the same tremendously weak arguments that kept on coming up again and again. The weakest argument in favour of AI was the perenial:
* Moore's Law h... |
a59a8efc-ee4e-458f-a648-d66ab6e869df | trentmkelly/LessWrong-43k | LessWrong | AI romantic partners will harm society if they go unregulated
Recently, when people refer to “immediate societal harms and dangers” of AI, in media or political rhetoric, they predominantly choose to mention “bias”, “misinformation”, and “political (election) manipulation”.
Despite politicians, journalists, and exper... |
77b502ad-48d3-413a-b948-a447f0531106 | trentmkelly/LessWrong-43k | LessWrong | In-Context Learning: An Alignment Survey
Epistemic status: new to alignment; some background. I learned about alignment about 1.5 years ago and spent the last ~1 year getting up to speed on alignment through 12 AI safety-related courses and programmes while completing an artificial intelligence MSc programme. Ultimate... |
346fcbdd-d8f8-48bf-acdc-297117f219d1 | trentmkelly/LessWrong-43k | LessWrong | How to Become a Magician
Crossposted from Nicole Dieker Dot Com, naturally. I wrote this one before I started following the LW community, so "magician" can be replaced with "hero" if you want — although I like using "magician" for reasons that will become clear in the post.
Start by asking yourself what you ought to ... |
6f5e85d9-f069-4346-8bff-b250637206d4 | trentmkelly/LessWrong-43k | LessWrong | Skirting the mere addition paradox
Consider the following facts:
1. For any population of people of happiness h, you can add more people of happiness less than h, and still improve things.
2. For any population of people, you can spread people's happiness in a more egalitarian way, while keeping the same average ha... |
112be0af-ad60-41b3-b824-c6cdea92170a | StampyAI/alignment-research-dataset/blogs | Blogs | Empowerment and Stakeholder Management
*Click lower right to download or find on Apple Podcasts, Spotify, Stitcher, etc.*
I've been writing recently about sweeping, very-long-run trends in whether things are going well or poorly in the world.[1](https://www.cold-takes.com/p/b38f340e-4b80-4823-adc5-0176ba638a4c/#fn1... |
39e995f9-9600-40da-8d9e-b62637f6b347 | trentmkelly/LessWrong-43k | LessWrong | The Study of Anglophysics
I.
Dear Dr. McCord:
Seven years ago, our research staff read with interest your work on Berkeleyan idealism. We were particularly fascinated by your seemingly outrageous claim that it might be possible for individuals to imagine mental worlds so strongly that they would take on a reality of... |
255c74aa-9120-4553-8bc5-ade162229e58 | trentmkelly/LessWrong-43k | LessWrong | Detection of Asymptomatically Spreading Pathogens
Cross-posted from my NAO Notebook.
This is an edited transcript of a talk I just gave at CBD S&T, a chem-bio defence conference. I needed to submit the slides several months in advance, so I tried out a new-to-me approach where the slides are visual support only and I... |
376275b3-55f9-4564-8d09-b6a95b2d7dbb | LDJnr/LessWrong-Amplify-Instruct | LessWrong | ""No! Try not! Do, or do not. There is no try." —Yoda Years ago, I thought this was yet another example of Deep Wisdom that is actually quite stupid. SUCCEED is not a primitive action. You can't just decide to win by choosing hard enough. There is never a plan that works with probability 1.
But Yoda was wiser than I fi... |
48c9531a-2d8b-4f5d-870c-54a18eeb8877 | trentmkelly/LessWrong-43k | LessWrong | Why do misalignment risks increase as AIs get more capable?
It's generally agreed that as AIs get more capable, risks from misalignment increase. But there are a few different mechanisms by which more capable models are riskier, and distinguishing between those mechanisms is important when estimating the misalignment ... |
9423d299-c5b3-482b-8f70-8aeaa542d308 | StampyAI/alignment-research-dataset/blogs | Blogs | MIRI’s recent effective altruism talks
MIRI recently participated in the 2014 Effective Altruism Retreat and [Effective Altruism Summit](http://www.effectivealtruismsummit.com/) organized by Leverage Research. We gave four talks, part... |
7488038e-20b6-4bb6-be10-f32edbcef3a7 | trentmkelly/LessWrong-43k | LessWrong | A quick calculation on exercise
The question is - am I doing enough exercise?
I intend to provide a worked example for you to work alongside with your own calculations and decide if you should increase or decrease your exercise.
The benefits of physical activity are various and this calculation can be done for one o... |
e4a45715-59bb-45e7-b23a-9bf58f112dca | trentmkelly/LessWrong-43k | LessWrong | A Brief Theology of D&D
At the beginning of history, entities gained power by eating each other. They each had goals, but few could accomplish anything in that environment. Over time the survivors gained information about each other, eventually allowing cooperation. They agreed to change the system: From now on, entit... |
8f4589b6-5e40-4492-9159-38a6790e6352 | trentmkelly/LessWrong-43k | LessWrong | Limitations of Laplace’s rule of succession
Summary
Laplace’s rule typically overestimates the chance of unprecedented events occurring; this is because it assigns 50% to the event occurring in the first timestep. We can do better by using a rule that assigns a lower probability to the event occurring in the timestep,... |
5c4b3ed2-7ef7-4eff-ba2e-ac03b07b4d0e | trentmkelly/LessWrong-43k | LessWrong | An observation about Hubinger et al.'s framework for learned optimization
The observations I make here have little consequence from the point of view of solving the alignment problem. If anything, they merely highlight the essential nature of the inner alignment problem. I will reject the idea that robust alignment, i... |
2ed28a47-7d77-409f-986a-fe73cd04b033 | trentmkelly/LessWrong-43k | LessWrong | More GPT-3 and symbol grounding
Based on work done with Rebecca Gorman and Oliver Daniel-Koch.
In a previous post, I talked about GPT-3 and symbol grounding. This post presents a simpler example where GPT-3 fails (and succeeds) at grounding its symbols.
Undead workers
The following text was presented to the Open AI... |
0f8b5cbb-137d-4f5d-ae52-e14907949871 | StampyAI/alignment-research-dataset/blogs | Blogs | A new guide to MIRI’s research
[](http://intelligence.org/research-guide/)Nate Soares has written “[A Guide to MIRI’s Research](http://intelligence.org/research-guide/),” which outlines the main thrusts of MIRI’... |
752879ef-3fb0-4e9c-a2b9-08b3edc5cd56 | StampyAI/alignment-research-dataset/arbital | Arbital | Emulating digits
In general, given enough $n$-digits, you can emulate an $m$-digit, for any $m, n \in$ [$\mathbb N$](https://arbital.com/p/45h). If $m < n,$ you can emulate an $m$-digit using just one $n$-digit — in other words, you can use a [https://arbital.com/p/-42d](https://arbital.com/p/-42d) like a $7$-di... |
2f8a4b78-f671-4069-9cbf-1fce340420e5 | trentmkelly/LessWrong-43k | LessWrong | Towards Understanding Sycophancy in Language Models
TL;DR: We show sycophancy is a general behavior of RLHF’ed AI assistants in varied, free-form text-generation settings, extending previous results. Our experiments suggest these behaviors are likely driven in part by imperfections in human preferences–both humans and... |
df6465ee-2426-4018-889a-4df616a7f0dc | trentmkelly/LessWrong-43k | LessWrong | Book Review: The End of Average
> We all strive to be like the others - more exactly, like the others, but better. (Chapter 2)
Note: the quotes have been translated back from the French translation, so they probably slightly differ from the original text.
The End of Average is a 250-ish pages book published in 20... |
9d928786-598d-4fbd-a951-114584285d43 | trentmkelly/LessWrong-43k | LessWrong | Basic Facts Beanbag
Summary: Take a beanbag. Pick a topic of simple information, such as the one digit multiplication tables or states and capitals or Scott’s Basic Numbers To Have Intuitions About Other Things. Toss the beanbag to someone else, asking a question as you do so. They catch it and answer as fast as they ... |
960fb38e-c358-463e-820a-2fa432ec11fe | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Logical Representation of Causal Models
*Epistemic status: I expect some people to say "this is obvious and trivial", and others to say "this makes no sense at all".*
One fundamental difference between [E.T. Jaynes’ probability textbook](https://www.amazon.com/Probability-Theory-Science-T-Jaynes/dp/0521592712) and mo... |
f9531a3c-6f25-4371-af46-f53610701b0c | trentmkelly/LessWrong-43k | LessWrong | is it possible to comment anonymously on a post?
|
e292968c-47f6-4af0-826b-60e4760cec17 | trentmkelly/LessWrong-43k | LessWrong | How much to spend on a high-variance option?
So the jackpot in the Ohio lottery is around 25 million, and the chance of winning it is one in roughly 14 million, with tickets at 1 dollar a piece. It appears to me that roughly a quarter million tickets are sold each drawing; so, supposing you win, the probability of som... |
1f6f23ab-604c-44c1-aa3b-0dfbcf6e303c | trentmkelly/LessWrong-43k | LessWrong | Papers on protein design
I've been catching up with recent progress and I wanted to share really brief summaries of the results I found interesting.
Training (mainly) on aminoacid sequences
* Large language models generate functional protein sequences across diverse families (Jul'21, Salesforce, 43 citations) allow... |
2fa3e2bd-e896-46fa-87a7-9e2bb9016ced | trentmkelly/LessWrong-43k | LessWrong | Conceptual issues in AI safety: the paradigmatic gap
None |
07f2a0bd-664e-4d69-98c0-cbe5880cb6f1 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Announcing Web-TAISU, May 13-17
**I am excited to announce Web-TAISU!**
**May 13-17, 2020**
I was going to run a [Technical AI Safety Unconference](https://www.lesswrong.com/events/BPTzfeQeZZ6chHvtr/taisu-technical-ai-safety-unconference-1) (TAISU) at [CEEALAR](https://ceealar.org/) (formerly EA Hotel), Blackpool, ... |
7c79724e-83d0-4911-b765-a906e771ed4e | trentmkelly/LessWrong-43k | LessWrong | AGI & Consciousness - Joscha Bach
In this post, I cover Joscha Bach' views on consciousness, how it relates to intelligence, and what role it can play to get us closer to AGI. The post is divided into three parts, first I try to cover why Joscha is interested in understanding consciousness, next, I go over what consc... |
1d79f70c-4da2-45b4-9bfe-cdfa3df502d5 | trentmkelly/LessWrong-43k | LessWrong | Where are other places to meet rationality-minded people?
This thread can be seen as a continuation of a previous thread (http://lesswrong.com/r/discussion/lw/33u/theoretical_target_audience_size_of_less_wrong/). I felt the need to make a new thread since this post is long and also since it is more targeted. This may ... |
ceaec38e-2537-4fce-9ec5-16a2e3ee4a8f | trentmkelly/LessWrong-43k | LessWrong | Survey of 2,778 AI authors: six parts in pictures
Crossposted from AI Impacts blog
The 2023 Expert Survey on Progress in AI is out, this time with 2778 participants from six top AI venues (up from about 700 and two in the 2022 ESPAI), making it probably the biggest ever survey of AI researchers.
People answered in O... |
e64a5ed2-3d83-4120-9d2a-c8b30ff5898b | trentmkelly/LessWrong-43k | LessWrong | Cooperation for AI safety must transcend geopolitical interference
> Amid a tech explosion, when we need to mobilise all our wisdom and energy for cooperation, some countries are trying to shut down collaborative platforms |
8662c93a-cb02-48c6-af56-81d1b0234138 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Gemini modeling
*[Metadata: crossposted from <https://tsvibt.blogspot.com/2022/08/gemini-modeling.html>. First completed 17 June 2022. I'm likely to not respond to comments promptly.]*
A gemini model is a kind of model that's especially relevant for minds modeling minds.
Two scenarios
=============
* You stand b... |
f32dbee3-6308-4efc-ae01-2ccaf9b24be5 | trentmkelly/LessWrong-43k | LessWrong | Perceptrons Explained
This is an introduction to the perceptron learning algorithm, a proof of its convergence, and some visualizations. Ideally, in the same vein as Nicky Case's Explorable Explanations, but I don't think it's at that level yet. Only LW-tangential, mostly just an educational post into basic ML. |
bb8993a8-17c3-4d0c-a4a6-3fa6df593484 | trentmkelly/LessWrong-43k | LessWrong | 2015 Repository Reruns - Boring Advice Repository
This is the first post of the 2015 repository rerun, which appears to be a good idea. The motivation for this rerun is that while the 12 repositories (go look them up, they're awesome!) exist and people might look them up, few new comments are posted there. In effec... |
f2362c64-7829-4306-b18d-75ceadc283d8 | trentmkelly/LessWrong-43k | LessWrong | Learning magic
Magic (primarily misdirection and cold-reading, but also the mechanics like sleight-of-hand) seems like an extremely good case-study in the study of how the human mind works and the predictable ways in which human maps differ from the territory. There are several magicians out there who offer to teach c... |
4ee37a67-e923-4c91-bd19-4b83fa3183cb | StampyAI/alignment-research-dataset/blogs | Blogs | January 2013 Newsletter
**Greetings from the Executive Director**
---
[](http://intelligence.org/blog/2012/11/07/november-2012-newsletter/luke/)
Dear friends of the Machine Intelligence Research Institu... |
0b62fe53-2190-46a1-add1-1e1e7b1b339e | trentmkelly/LessWrong-43k | LessWrong | Some negative steganography results
Disclaimer: These are negative results and this post is below the normal quality bar I usually use. These were quick experiments (a couple of days of work) and I don’t have much expertise in RL. They don’t provide strong evidence about the natural emergence of steganography.
People... |
05bc62ab-d34d-4bcf-a086-ca50759e8caf | trentmkelly/LessWrong-43k | LessWrong | 2016 LessWrong Diaspora Survey Analysis: Part Two (LessWrong Use, Successorship, Diaspora)
2016 LessWrong Diaspora Survey Analysis
Overview
* Results and Dataset
* Meta
* Demographics
* LessWrong Usage and Experience
* LessWrong Criticism and Successorship
* Diaspora Community Analysis (You are here)
* Mental... |
334b5764-9b2e-44e1-a44b-d9a8357e50cb | StampyAI/alignment-research-dataset/arxiv | Arxiv | Training language models to follow instructions with human feedback
1 Introduction
---------------
Large language models (LMs) can be “prompted” to perform a range of natural language processing (NLP) tasks, given some examples of the task as input. However, these models often express unintended behaviors such as m... |
d066b904-48c8-4933-9f50-51fe269f1fed | trentmkelly/LessWrong-43k | LessWrong | Welcome to Analogia! (Chapter 7)
Hey, what's that sound?
"It's time we stop
Hey, what's that sound?
Everybody look, what's going down?
Paranoia strikes deep
Into your life it will creep
It starts when you're always afraid
Step out of line, the men come and take you away
We better stop
Hey, what's that sound?
Everybody... |
840a2ec7-f82f-44f2-9778-bf9e01e03c11 | trentmkelly/LessWrong-43k | LessWrong | Can solipsism be disproven?
I had a discussion with a find of mine via text messaging on whether solophisim can be proved or disproved. The conversation went a somewhat off track, but I believe it produced some conclusions that would be of interest to others. I shall refer to my friend as Thomas Aquinas. The conve... |
9b641f58-2c17-49ef-894c-87b2a93f146f | trentmkelly/LessWrong-43k | LessWrong | An argument against excessive consumption of coffee from 1674
One every now and then discovers the most amusing writings buried in the dusty and forgotten past. The following was apparently in response to the dramatic increase in coffee consumption of renaissance England, from an eccentric’s curiosity to a widespread ... |
e75de094-6eea-43c0-a5c6-e210a754d759 | trentmkelly/LessWrong-43k | LessWrong | If you want to learn technical AI safety, here's a list of AI safety courses, reading lists, and resources
If you're interested in technical AI safety, it's hard to know where to start. Lots of people have made courses and reading lists to help simplify things.
We made a spreadsheet of such resources for learning ab... |
f3bb73a7-0dec-4c9e-8479-cf6844cebff1 | trentmkelly/LessWrong-43k | LessWrong | Daniel Kahneman on Charlie Rose [video]
Daniel Kahneman (of Tversky and Kahneman fame) was interviewed on PBS's Charlie Rose last night, discussing his book Thinking, Fast and Slow (reviewed here by lukeprog):
http://www.charlierose.com/view/interview/12185 |
a727d55f-73a2-4866-93e0-1cc533b1ef6e | trentmkelly/LessWrong-43k | LessWrong | Morality is Accidental & Self-Congratulatory
The Selfish Gene remains one of my favorite books of all time and although published in 1976 it remains a compelling and insightful introduction to the brute mechanics of natural selection. Richard Dawkins later acknowledged that his book’s title may give a misleading impre... |
12c59c77-41ff-441c-836a-66c43fe26c28 | trentmkelly/LessWrong-43k | LessWrong | Improving My Writing Style
I've written quite a lot here since Less Wrong started up, but I've started to suspect that my writing style is holding me back. Most recently, I wrote two sequences that seemed to garner widespread agreement on content/significance/originality but didn't really seem to excite anyone, which ... |
77e78836-62d2-404c-8873-ee5f74fb159e | trentmkelly/LessWrong-43k | LessWrong | Meetup : Atlanta Movie Night: Introduction to Machine Learning
Discussion article for the meetup : Atlanta Movie Night: Introduction to Machine Learning
WHEN: 22 June 2014 07:00:35PM (-0400)
WHERE: 2388 Lawrenceville Hwy. Unit L Decatur, GA 30033
Stanford professor, Andrew Ng, is teaching a course on Coursera about... |
2e55466f-0fb8-4b1f-9515-b4736f3f04e5 | trentmkelly/LessWrong-43k | LessWrong | List of machine learning researchers
This is not a long post, but I wanted to draw attention to Wikipedia's list of notable machine learning researchers, because it has done so much for me. For anyone interested in problems related to AI (to get a feel for what AI is like), I recommend reading at least the Wikipedia p... |
06a9812d-df16-4b75-a809-1b6016b3e985 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post616
In this post, we analyze some simple games that model the problem of ensuring AIs never succeed at taking a single catastrophic action (which we'll refer to as "escaping") during deployment. This exercise highlights some of the difficulties associated with deploying misaligned models given a limited perfec... |
f314e821-7613-41cc-a710-22a82b1e2d0d | trentmkelly/LessWrong-43k | LessWrong | Test Your Rationality
So you think you want to be rational, to believe what is true even when sirens tempt you? Great, get to work; there's lots you can do. Do you want to justifiably believe that you are more rational than others, smugly knowing your beliefs are more accurate? Hold on; this is hard.
Humans nearl... |
7c88498b-dc2c-423b-a8cc-aada5915a076 | trentmkelly/LessWrong-43k | LessWrong | The Alignment Newsletter #7: 05/21/18
Highlights
Challenges to Christiano’s capability amplification proposal (Eliezer Yudkowsky): A list of challenges faced by iterated distillation and amplification. First, a collection of aligned agents interacting does not necessarily lead to aligned behavior. (Paul's response: Th... |
fcfd23ad-d9c5-45d3-b24b-c6da6aa42079 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Challenge: construct a Gradient Hacker
This is a relatively clean subproblem that we came upon a few months ago while thinking about gradient hacking. We're throwing it out to the world to see if anyone can make progress.
**Problem:** Construct a gradient hacker (definition below), or prove that one cannot exist und... |
df679be4-aaa9-48b7-a750-28dea6df4dc0 | trentmkelly/LessWrong-43k | LessWrong | Weekly LW Meetups
This summary was posted to LW main on April 25th. The following week's summary is here.
New meetups (or meetups with a hiatus of more than a year) are happening in:
* Christchurch, NZ Inaugural Meetup: 27 April 2014 04:30PM
* Las Vegas/Henderson, NV: 01 May 2014 06:00PM
Irregularly scheduled Les... |
4e46eea0-423d-429c-ae78-ea85a6dc9f87 | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | DeepMind: The Podcast (S2 trailer)
i'm professor hannah fry a mathematician
and podcaster and i'm excited to tell
you about the return of deepmind the
podcast
it's the series where we go behind the
scenes of one of the world's leading
artificial intelligence companies
to find out how the latest developments
in ai
are ... |
00523ae6-ef65-4760-8758-4993c15534b3 | trentmkelly/LessWrong-43k | LessWrong | Open Thread June 2010, Part 4
This thread is for the discussion of Less Wrong topics that have not appeared in recent posts. If a discussion gets unwieldy, celebrate by turning it into a top-level post.
This thread brought to you by quantum immortality. |
01a6d846-44be-420d-bbf8-4ce0410477b5 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Monthly Ohio meetup
Discussion article for the meetup : Monthly Ohio meetup
WHEN: 22 July 2012 04:00:00PM (-0400)
WHERE: The Dublin Pub, 300 Wayne Avenue, Dayton
Our usual monthly meetup; note the unusual date and place. This month's topic is the next set of minicamp exercises. Erica, Elizabeth, Alex, and R... |
5eed92be-caaa-4802-bc08-4a00a4ec007c | StampyAI/alignment-research-dataset/special_docs | Other | Life 3.0: Being Human in the Age of Artificial Intelligence (2017, Alfred A. Knopf)
[]
Also by Max Tegmark
Our Mathematical Universe
[Life 3.0 Being Human in the Age of Artificial Intelligence Max Tegmark Alfred A. Knopf New York 2017][Life 3.0 Being Human in the Age of Artificial Intelligence Max Tegmark Alfred A.... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.