id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
3cd480d6-bbb1-4566-bcfe-365d9ce044a9
trentmkelly/LessWrong-43k
LessWrong
Bias in rationality is much worse than noise Crossposted at Intelligent Agent Forum. I've found quite a few people dubious of my "radical skepticism" post on human preferences. Most "radical skepticism" arguments - Descartes's Demon, various no-free-lunch theorems, Boltzmann Brains - generally turn out to be irreleva...
d9b45243-1acb-4c91-96b1-f72612194a9a
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Unfaithful Explanations in Chain-of-Thought Prompting Introduction ------------ I recently released “[Language Models Don't Always Say What They Think: Unfaithful Explanations in Chain-of-Thought Prompting](https://arxiv.org/abs/2305.04388)” with collaborators Julian Michael, Ethan Perez, and Sam Bowman. For a summar...
10bae926-ae42-4baa-a9ea-dacb56716a8c
trentmkelly/LessWrong-43k
LessWrong
Extensions and Intensions > "What is red?" > "Red is a color." > "What's a color?" > "A color is a property of a thing." But what is a thing?  And what's a property?  Soon the two are lost in a maze of words defined in other words, the problem that Steven Harnad once described as trying to learn Chinese from a Chines...
a24d591b-6c49-45d0-94a8-ac8c5c8b8438
trentmkelly/LessWrong-43k
LessWrong
Will early transformative AIs primarily use text? [Manifold question] Will early transformative AIs primarily use text to store intermediate thoughts and memories? I think the answer to this question matters for alignment research prioritization, and getting some common knowledge about what are the most commonly held...
b7462298-7889-4de0-aee9-3235b40b0c95
trentmkelly/LessWrong-43k
LessWrong
Deception Chess: Game #1 (You can sign up here if you haven't already.) This is the first of my analyses of the deception chess games. The introduction will describe the setup of the game, and the conclusion will sum up what happened in general terms; the rest of the post will mostly be chess analysis and skippable i...
720d5f6d-b97e-4bb0-9f52-94430e3bde8b
trentmkelly/LessWrong-43k
LessWrong
What the cost difference in processing input vs. output tokens with LLMs? Hi, I am trying to understand the difference in the cost of producing a single input token vs output token. Based on some articles, I came to the following conclusion: 1. Input tokens processing scales quadratically, there’s no way around it...
3b646d25-c0e5-429e-a972-3dc08be8fe5e
trentmkelly/LessWrong-43k
LessWrong
Machines of Stolen Grace TLDR: Initial RL training runs (like those contributing to o3's capabilities) give rise to basic reasoning heuristics (perhaps forming nascent reasoning circuits) that intelligently mimic patterns in the training data. Massively scaling this RL on larger base models presents a potential pathwa...
6d3275c3-e142-42c7-bb05-073c33b68c66
trentmkelly/LessWrong-43k
LessWrong
LessWrong analytics (February 2009 to January 2017) Table of contents * Introduction * Pageviews and sessions * Top posts * Source code * Further reading * Acknowledgments Introduction In January 2017, Vipul Naik obtained Google Analytics daily sessions and pageviews data for LessWrong from Kaj Sotala. Vipul a...
e52dc6d0-7cd7-49e4-9dcf-f45496b34d84
trentmkelly/LessWrong-43k
LessWrong
The argument for near-term human disempowerment through AI Author: Leonard Dung Abstract: Many researchers and intellectuals warn about extreme risks from artificial intelligence. However, these warnings typically came without systematic arguments in support. This paper provides an argument that AI will lead to the p...
4442e0e1-265a-4840-a59d-aba54fc3f70f
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: cbc1b5acada36b0f There are two fundamentally different constraints on schemers Redwood Research blog Subscribe Sign in Share this post Redwood Research blog There are two fundamentally different constraints on schemers Copy link Facebook Email Notes More There are two fundamentally different constraints on schemer...
e2cdfc11-9fbd-4d08-a724-41a5b6d35b6b
trentmkelly/LessWrong-43k
LessWrong
Chemical Turing Machines Epistemic status: brief writeup of some interesting work I found in[1][2][3][4], among other places. Probably subtly wrong in places & if you have additions / comments to make, please do. Finite state automata (FSA) can be modeled with reactions of the form A+B→C+D. FSAs operate over regular ...
cbd064d9-c021-46b5-ba09-42beb62e7ce4
StampyAI/alignment-research-dataset/arxiv
Arxiv
Can You Trust Your Model's Uncertainty? Evaluating Predictive Uncertainty Under Dataset Shift 1 Introduction --------------- Recent successes across a variety of domains have led to the widespread deployment of deep neural networks (DNNs) in practice. Consequently, the predictive distributions of these models are i...
79a0ce5b-5d54-445d-b47e-dcee69cf03b8
trentmkelly/LessWrong-43k
LessWrong
The law of effect, randomization and Newcomb’s problem None
638fd840-88fb-4a6b-a317-bfd2eb1ec17a
trentmkelly/LessWrong-43k
LessWrong
Truth and Advantage: Response to a draft of "AI safety seems hard to measure" Status: This was a response to a draft of Holden's cold take "AI safety seems hard to measure". It sparked a further discussion, that Holden recently posted a summary of. The follow-up discussion ended up focusing on some issues in AI align...
cbbe87f2-0d40-4e77-97dc-7252566532db
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Human instincts, symbol grounding, and the blank-slate neocortex Intro: What is Common Cortical Algorithm (CCA) theory, and why does it matter for AGI? ====================================================================================== As I discussed at [Jeff Hawkins on neuromorphic AGI within 20 years](https://w...
d0bc23e6-826a-4262-ad40-7d1658014762
trentmkelly/LessWrong-43k
LessWrong
Updating on Bad Arguments Here is an intuitively compelling principle: hearing a bad argument for a view shouldn’t change your degree of belief in the view. After all, it is possible for bad arguments to be offered for anything, even the truth. For all you know, plenty of good arguments exist, and you just happened to...
daed4238-a9e8-41d2-9607-cb64d09faa57
trentmkelly/LessWrong-43k
LessWrong
Turing-Test-Passing AI implies Aligned AI Summary: From the assumption of the existence of AIs that can pass the Strong Form of the Turing Test, we can provide a recipe for provably aligned/friendly superintelligence based on large organizations of human-equivalent AIs > Turing Test (Strong Form): for any human H th...
3f165116-d4a8-4d8b-a86a-0292ea48360a
trentmkelly/LessWrong-43k
LessWrong
Discrimination: less is more ‘Discrimination’ can mean all sorts of things. One of the main ones, and what it will mean in this post, is differential treatment of people from different groups, due to real or imagined differences in average group features.  Discrimination is a problem because the many people who don’t ...
b1be65c2-ea67-4ede-bd58-4c551ad2d7ad
trentmkelly/LessWrong-43k
LessWrong
Sunk Cost Fallacy Related to: Just Lose Hope Already, The Allais Paradox, Cached Selves In economics we have this concept of sunk costs, referring to costs that have already been incurred, but which cannot be recouped. Sunk cost fallacy refers to the fallacy of honoring sunk costs, which decision-theoretically should...
f9c445a1-2ab8-4f61-9c0f-957334a1e945
StampyAI/alignment-research-dataset/special_docs
Other
Preference learning along multiple criteria: A game-theoretic perspective. Preference learning along multiple criteria: A game-theoretic perspective Kush Bhatia EECS, UC Berkeley kush@cs.berkeley.eduAshwin Pananjady Simons Institute, UC Berkeley ashwinpm@berkeley.edu Peter L. Bartlett EECS and Statistics, UC Berkeley ...
025989a7-8561-4881-bc8f-7aaed25cfe54
trentmkelly/LessWrong-43k
LessWrong
Depression as a defense mechanism against slavery Individuals who have to conform to an external authority too powerful to resist often get depressed: which among other effects, includes lower productivity and higher risk for suicide.  This suggests one way how a tendency for depression, and resulting behaviors of akr...
40dcc040-640a-4d16-986b-416966f6c5b9
trentmkelly/LessWrong-43k
LessWrong
Open thread, Feb. 06 - Feb. 12, 2017 If it's worth saying, but not worth its own post, then it goes here. ---------------------------------------- Notes for future OT posters: 1. Please add the 'open_thread' tag. 2. Check if there is an active Open Thread before posting a new one. (Immediately before; refresh the ...
38469de1-1647-4dc0-9999-a2ae03bcb4f1
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
How could a moratorium fail? The debates this week have clarified a number of things in my mind, and I think that has been useful. At the same time, I think there’s a lack of clarity about what was proposed, and what the objections are. Given that, I wanted to summarize my position, and then, despite being convinced t...
32f8ab0b-b837-4cfc-8189-11bfb7048fd0
trentmkelly/LessWrong-43k
LessWrong
what an efficient market feels from inside > “I often think of the time I met Scott Sumner and he said he pretty much assumes the market is efficient and just buys the most expensive brand of everything in the grocery store.” - a Tweet It’s a funny quip, but it captures the vibe a lot of people have about efficient m...
0e1ddd1d-af3a-4c7e-996c-498e13f9fb90
trentmkelly/LessWrong-43k
LessWrong
Evaluations (of new AI Safety researchers) can be noisy TL;DR: Evaluating whether or not someone will do well at a job is hard, and evaluating whether or not someone has the potential to be a great AI safety researcher is even harder. This applies to evaluations from other people (e.g. job interviews, first impression...
b60341f9-c4dc-4f21-b0d7-1bdab7dea353
trentmkelly/LessWrong-43k
LessWrong
“Who’s In Charge? Free Will and the Science of the Brain” If everything, including you and me, is made up of material that blindly obeys the inflexible laws of physics, then everything that happens, including what you and I do, is inevitable, and free will is something of an illusion or a joke. Right? Or are we just...
3acea5bb-3fe6-4017-8ab0-9cb86b623dbe
trentmkelly/LessWrong-43k
LessWrong
A new intro to Quantum Physics, with the math fixed Confidence level: I am a physicist with a phd in computational quantum chemistry. I’m pretty sure there are no major errors here, but I still may have missed something. Thanks to Blake Stacey for looking over the post. The goal of this post is to rewrite everything ...
6b3bcaed-b47c-4a57-a40b-0f32df86edbf
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Akrasia is the tendency to act against your own long-term interests, and is a problem doubtless only too familiar to us all. In his book "Breakdown of Will", psychologist George C Ainslie sets out a theory of how akrasia arises and why we do the things we do to fight it. His extraordinary proposal takes insights given...
6e72b7f3-3cad-44a0-bba9-5ddee2ef236c
trentmkelly/LessWrong-43k
LessWrong
[Stub] The wrong philosophical question Guy who wants to do philosophy with me: "If you made a perfect copy of yourself and destroyed the original, would that copy be you?" Me: "Oh yes, of course. Trivially." GWWTDPWM: "Ah, but what if you didn't destroy the original immediately, and then..." Me: "What? You have tw...
2d0d9203-dc32-45ef-b6dd-17b15d1b2c26
trentmkelly/LessWrong-43k
LessWrong
An update on Signal Data Science (an intensive data science training program) In December 2015, Robert Cordwell and I cofounded Signal Data Science (website), which we announced on Less Wrong. Our first cohort has just concluded, and overall went very well. We're planning another one in Berkeley from May 2nd – June ...
7d72cbea-06cc-4b3c-8a8f-fce6dbdaa76e
trentmkelly/LessWrong-43k
LessWrong
Meetup : West LA—How to Live on 22 Hours a Day Discussion article for the meetup : West LA—How to Live on 22 Hours a Day WHEN: 08 January 2014 07:00:00PM (-0800) WHERE: 10850 West Pico Blvd, Los Angeles, CA 90064, USA How to get in: Go to the Westside Tavern in the upstairs Wine Bar (all ages welcome), located insi...
15064277-6de0-4127-9cf2-6f34f608a0f4
trentmkelly/LessWrong-43k
LessWrong
NYT: A.I. Researchers Are Making More Than $1 Million, Even at a Nonprofit None
ec68c485-f967-43f8-abce-383da38cc406
trentmkelly/LessWrong-43k
LessWrong
Smoke without fire is scary This post was written in response to Evan Hubinger’s shortform prompt below and benefited from discussions with him. Thanks to Justis Mills for feedback on this post. > If we see precursors to deception (e.g. non-myopia, self-awareness, etc.) but suspiciously don’t see deception itself, th...
ae108777-e85b-429d-afad-518c16646369
trentmkelly/LessWrong-43k
LessWrong
Sam Harris' surprisingly modest proposal Sam Harris has a new book, The Moral Landscape, in which he makes a very simple argument, at least when you express it in the terms we tend to use on LW: he says that a reasonable definition of moral behavior can (theoretically) be derived from our utility functions. Essentiall...
bde5d7a6-2605-4e62-862d-0b29ffb523a7
trentmkelly/LessWrong-43k
LessWrong
The Great Filter isn't magic either Crossposted at Less Wrong 2.0. A post suggested by James Miller's presentation at the Existential Risk to Humanity conference in Gothenburg. Seeing the emptiness of the night sky, we can dwell upon the Fermi paradox: where are all the alien civilizations that simple probability est...
0eaa6108-c87c-4135-9420-e9bcd440dbe7
StampyAI/alignment-research-dataset/arxiv
Arxiv
Path-Specific Objectives for Safer Agent Incentives 1 Introduction --------------- Artificial agents can have unsafe incentives to influence parts of their environments in unintended ways. For example, content recommendation systems can achieve good performance by manipulating their users to develop more predictabl...
3eebfb05-032e-4082-b8da-9fbaba394764
StampyAI/alignment-research-dataset/arxiv
Arxiv
Universal Algorithmic Intelligence: A mathematical top->down approach 1 Introduction --------------- This article gives an introduction to a mathematical theory for intelligence. We present the AIXI model, a parameter-free optimal reinforcement learning agent embedded in an arbitrary unknown environment. The sc...
7f8cfa0d-82de-4077-bc1b-76edbdb7402e
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Relevant pre-AGI possibilities > The default, when reasoning about advanced artificial general intelligence (AGI), is to imagine it appearing in a world that is basically like the present. Yet almost everyone agrees the world will likely be importantly different by the time advanced AGI arrives. > One way to address...
15b98319-df1d-4dd4-b775-f40c52fd99e3
StampyAI/alignment-research-dataset/lesswrong
LessWrong
AI scares and changing public beliefs The opportunity --------------- I, for one, am looking forward to the the next public AI scares. There's a curious up-side to the reckless deployment of powerful, complex systems: they scare people. This has the potential to shift the landscape of the public debate. It's an oppo...
542b9fbe-133e-4997-8bb6-52cd68280846
StampyAI/alignment-research-dataset/arxiv
Arxiv
Semantic Image Synthesis with Spatially-Adaptive Normalization 1 Introduction --------------- Conditional image synthesis refers to the task of generating photorealistic images conditioning on some input data. Earlier methods compute the output image by stitching pieces from a database of images [[13](#bib.bib13), ...
db38c7c3-65ca-44be-b98e-50619ef55cf0
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
EA Infosec: skill up in or make a transition to infosec via this book club Ahoy! Our community has become acutely aware of the need for skilled infosec folks to help out in all cause areas. The market conditions are that information security skilled individuals are in shorter supply than demand. This book club aims to...
fed6cba3-7e0f-4c2f-aeee-3f3f7d39a8fe
trentmkelly/LessWrong-43k
LessWrong
You Are Not Measuring What You Think You Are Measuring Eight years ago, I worked as a data scientist at a startup, and we wanted to optimize our sign-up flow. We A/B tested lots of different changes, and occasionally found something which would boost (or reduce) click-through rates by 10% or so. Then one week I was p...
5fd484ef-6da1-4fd6-b138-52d97eabc73c
trentmkelly/LessWrong-43k
LessWrong
A failed attempt at Updatelessness using Universal Inductors Here, I present a failed attempt to build an updateless decision theory out of universal inductors. It fails because it is is mistaking updatelessness about which logical theory it is in for true logical updatelessness about computations. I will use Tsvi's n...
d9055a24-d149-4bb7-a9b5-df61c8dfcce3
trentmkelly/LessWrong-43k
LessWrong
Meetup : Israel Less Wrong Meetup - Social and Board Games Discussion article for the meetup : Israel Less Wrong Meetup - Social and Board Games WHEN: 03 April 2014 07:00:00PM (+0300) WHERE: Google Tel Aviv We're going to have a meetup on Thursday, April 3rd at Google Israel's offices, Electra Tower, 98 Yigal Alon ...
1cb57352-1ab7-4a77-a013-610df747251d
StampyAI/alignment-research-dataset/arxiv
Arxiv
A Foundry of Human Activities and Infrastructures A Foundry of Human Activities and Infrastructures Robert B. Allen, Eunsang Y ang, and Tat sawan Timakum Y onsei University, Seoul, South Korea rballen@yonsei.ac.kr, esy220@nyu.edu, tatsawan@gmail.com Abstract. Direct rep resentation knowledgebases can enhance ...
561707c3-694d-4a50-a363-281a49fe20af
trentmkelly/LessWrong-43k
LessWrong
Losing Metaphors: Zip and Paste In python (and several other languages) if I have two lists and want to process corresponding elements together I can use zip: >>> for number, letter in zip( ... [1,2,3,4], ["a", "b", "c", "d"]): ... print(number, letter) ... 1 a 2 b 3 c 4 d The metaphor is a zipper, taking the...
e71f4267-a49d-47fe-b275-1cfd77458b08
trentmkelly/LessWrong-43k
LessWrong
AISN #20: LLM Proliferation, AI Deception, and Continuing Drivers of AI Capabilities Welcome to the AI Safety Newsletter by the Center for AI Safety. We discuss developments in AI and AI safety. No technical background required Subscribe here to receive future versions. ---------------------------------------- AI ...
e485781b-f188-48fd-8b9d-9c791fd26594
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Late 2021 MIRI Conversations: AMA / Discussion With the release of [Rohin Shah and Eliezer Yudkowsky's conversation](https://www.lesswrong.com/posts/tcCxPLBrEXdxN5HCQ/shah-and-yudkowsky-on-alignment-plans), the **Late 2021 MIRI Conversations** sequence is now complete. This post is intended as a generalized comment s...
f492d7c4-7ff6-4892-8a54-053309336c4c
trentmkelly/LessWrong-43k
LessWrong
A most peculiar utility function In a previous post, I glibly suggested a utility function u# that would allow a u-maximising agent to print out the expectation of the utility it was maximising. But u# turns out to be very peculiar indeed. For this post, define * u#=−q2+2qu, for some q that is a future output of ...
6a468ed0-c2da-4cad-b7b6-b343c01529c6
trentmkelly/LessWrong-43k
LessWrong
Succeess depends on finding a balance between "geek" and "jock" Summary: title + LW seems to be already coming from a culture like that, I just want to make it conscious. Also, it is true for individuals and true for groups like nations. Why too much "jock" is bad Anti-intellectualism, black-and-white tribalism, imp...
24ce5521-47ad-4b50-be97-d0f31a25a79b
trentmkelly/LessWrong-43k
LessWrong
Superintelligence 22: Emulation modulation and institutional design This is part of a weekly reading group on Nick Bostrom's book, Superintelligence. For more information about the group, and an index of posts so far see the announcement post. For the schedule of future topics, see MIRI's reading guide. -------------...
84169af2-4f3f-41fe-ada4-e862e171f341
trentmkelly/LessWrong-43k
LessWrong
Covid 9/17: Done Biden His Time No more mister nice guy. The dude will not a-Biden. All this vaccine hesitancy is a bunch of malarkey. If you’re not vaccinated, the President is mad as hell and he’s not going to take it anymore.  The question then becomes, what is he gonna do about it, under what authority, and what ...
f19a14b8-a238-4ef1-9951-f3cd7c2c49a8
trentmkelly/LessWrong-43k
LessWrong
Asteroids and spaceships are kinetic bombs and how to prevent catastrophe A reality of physics, and one that doesn't get much play in science fiction, is that as soon as humanity gains space travel, anyone in the asteroid mining or space travel business will have city-busting capabilities at their fingertips. It's th...
0955e7a8-f9dd-46b9-b347-af9d2ca88172
trentmkelly/LessWrong-43k
LessWrong
The Chromatic Number of the Plane is at Least 5 - Aubrey de Grey This is a long standing open problem in math. Many people learn about it as early as high school as an example of an open problem, because it is so easy to state: How many colors does it take to color every point in the plane so that ever pair of points...
0fa30454-2c1b-470c-b43f-ccaa129182a4
trentmkelly/LessWrong-43k
LessWrong
AI #121 Part 2: The OpenAI Files You can find Part 1 here. This resumes the weekly, already in progress. The primary focus here is on the future, including policy and alignment, but also the other stuff typically in the back half like audio, and more near term issues like ChatGPT driving an increasing number of people...
a350ee18-2100-4390-9dc1-62d1c102feec
trentmkelly/LessWrong-43k
LessWrong
What are biases, anyway? Multiple type signatures With thanks to Rebecca Gorman for helping develop this idea. I've been constructing toy examples of agents with preferences and biases. There turns out to be many, many different ways of doing this, and none of the toy examples seem very universal. The reason for this...
8c3d3231-f771-4310-a7cc-b91fe9d7bb7d
StampyAI/alignment-research-dataset/arxiv
Arxiv
Adapting Language Models for Zero-shot Learning by Meta-tuning on Dataset and Prompt Collections 1 Introduction --------------- The goal of zero-shot classification (ZSC) is to classify textual inputs using label descriptions without any examples Yin et al. ([2019](#bib.bib59)). Large language models - whose only t...
46da764e-8f4b-4409-9c63-73518b605087
trentmkelly/LessWrong-43k
LessWrong
PSA: if you are in Russia, probably move out ASAP The Metaculus crowd thinks that it's very likely that Russia will restrict emigration:   I myself am more optimistic about this question resolving negatively. But even if Russia would not close borders for a notable fraction of its citizens, it seems fairly likely t...
da06bcc0-a1b6-48e4-993a-a4614b13098a
StampyAI/alignment-research-dataset/lesswrong
LessWrong
What is the ground reality of countries taking steps to recalibrate AI development towards Alignment first? You can possibly put the lid on European AI research, the biggest sign to this is civilian administration oversight over the internet which is incredibly tight compared to European Union's non authoritarian gove...
3915f06f-28f3-4591-a2f5-0eeb17a1d666
trentmkelly/LessWrong-43k
LessWrong
How to Support Someone Who is Struggling [Crossposted from my blog] There’s no shortage of pain, tragedy and loss in the world. And if you’re anything like me, you don’t always know how to be helpful when a loved one is going through the worst of it. Over the past few years, I’ve been trying to get better at that. ...
15ca107b-c0d0-442a-a394-7c259fe89332
trentmkelly/LessWrong-43k
LessWrong
Simplicity arguments for scheming (Section 4.3 of "Scheming AIs") This is Section 4.3 of my report “Scheming AIs: Will AIs fake alignment during training in order to get power?”. There’s also a summary of the full report here (audio here). The summary covers most of the main points and technical terms, and I’m hoping ...
8cc01f82-b114-45cc-92bb-c00a06283f09
trentmkelly/LessWrong-43k
LessWrong
My Favorite Productivity Blog Posts I’ve read at least a few hundred blog posts, maybe upwards of a thousand. Agreeing with Gavin Leech, I believe I’ve gained from essays more than any other medium. I’m an intellectually curious college student with a strong interest in both the practical and philosophical. This blog ...
dacd5a42-2594-49af-a0ee-32c5d0928f4b
trentmkelly/LessWrong-43k
LessWrong
Actors and scribes, words and deeds [Epistemic status: exploratory exercise in naming and concept-formation.] Among the kinds of people, are the Actors, and the Scribes. Actors mainly relate to speech as action that has effects. Scribes mainly relate to speech as a structured arrangement of pointers that have meaning...
83aaab78-3efa-405a-bae4-62d90d66ffed
trentmkelly/LessWrong-43k
LessWrong
Early career choices to maximise learning? I'm a student in a subfield of CS and biology, getting my bachelor's degree later this year and planning on continuing to masters (and maybe beyond). My ultimate goal is to work on the edge between biology, CS and datascience, preferably in an industry-backed research centre ...
6c3923eb-65f9-4940-b296-536e5f40a043
StampyAI/alignment-research-dataset/lesswrong
LessWrong
An Intuitive Introduction to Evidential Decision Theory In the [last post](https://www.lesswrong.com/posts/bNayfvnKKsbE7w6Sb/an-introduction-to-causal-decision-theory), we discussed Causal Decision Theory (CDT). We learned how purely looking at the causal effects of your actions makes you lose at Newcomb's Problem. [E...
cf748cf2-a50f-435f-a28a-db0d6d363cb9
trentmkelly/LessWrong-43k
LessWrong
Stop Making Sense Epistemic Status: Seven years and one day ago, I wrote this. In 2017 we faced years of Trump and couldn't imagine covid. Here in 2024 we face years of Trump and I'm trying to "imagine what we can't imagine". I dislike the title (except for the part where it might introduce people to music from a prev...
a0a57706-25ca-4748-8b8c-8ab40d6d0042
trentmkelly/LessWrong-43k
LessWrong
Commitment Races are a technical problem ASI can easily solve Current status (2025 April 16): Dang it, Commitment Races may remain a serious problem because I discovered a potentially fatal flaw in my argument. See my comment below. I still think the solution I outlined might work, but the reason it works no longer i...
03ccdcba-8644-4201-9abf-4493fbe3c0c3
trentmkelly/LessWrong-43k
LessWrong
Celebrate Trivial Impetuses There is a flipside to the trivial inconvenience: the trivial impetus.  This is the objectively inconsequential factor that gets you off your rear and doing something you probably would have left undone.  It doesn't have to be a major, crippling akrasia issue.  I'm not talking so much about...
9d0f3cab-5082-499b-a26c-7c175131ffcc
StampyAI/alignment-research-dataset/arxiv
Arxiv
Safety Assurances for Human-Robot Interaction via Confidence-aware Game-theoretic Human Models. I Introduction --------------- We focus on maintaining safety in highly dynamic human-robot interactions, such as when an autonomous car merges into a roundabout with an oncoming human-driven vehicle (Fig. [1](#S1.F1 "Fi...
0afcec5c-4bdd-4ea3-9406-9b80f2a10ff0
trentmkelly/LessWrong-43k
LessWrong
Help with relevant rational decision making research ideas Im researching rational medical decision making and trying to find right angles to approach the subject of probabilistic decision making in medicine. My first paper will be about the probability updating after the medical tests when we know the prevalence of...
1203b5e9-a1b7-4b69-967f-5b857f82019b
trentmkelly/LessWrong-43k
LessWrong
Tentative covid surface risk estimates My household previously made some highly uncertain estimates of the covid risk from bringing random objects that other people have recently been touching into our home, for instance salads and groceries an endless stream of Amazon packages. The official guidance is very vague, e....
c04d3a35-82d8-4647-82f1-d91e04ea9385
trentmkelly/LessWrong-43k
LessWrong
A naive alignment strategy and optimism about generalization (Context: my last post was trying to patch a certain naive strategy for AI alignment, but I didn’t articulate clearly what the naive strategy is. I think it’s worth explaining the naive strategy in its own post, even though it’s not a novel idea.) Suppose t...
dbdd9787-71ba-4e71-8b58-a883e442f5ba
trentmkelly/LessWrong-43k
LessWrong
Toy model of the AI control problem: animated version A few years back, I came up with a toy model of the AI control problem. It has a robot moving boxes into a hole, with a slightly different goal than it's human designers, and a security camera to check that it's behaving as it should. The robot learns to block the ...
5ad3765a-3cc9-4b52-8985-8d890e3821d0
trentmkelly/LessWrong-43k
LessWrong
Paper: Large Language Models Can Self-improve [Linkpost] This paper came out about a week ago. I am not the author - it was published anonymously. I learned about the paper when JJ Hepburn shared it on Slack. I just thought it seemed potentially really important and I hadn't seen it discussed on this forum yet: ----...
7df961d0-41b5-45e1-bbdb-3330b8c621cc
trentmkelly/LessWrong-43k
LessWrong
MIRI 2024 Mission and Strategy Update As we announced back in October, I have taken on the senior leadership role at MIRI as its CEO. It’s a big pair of shoes to fill, and an awesome responsibility that I’m honored to take on. There have been several changes at MIRI since our 2020 strategic update, so let’s get into ...
9ac333d0-4098-4f2d-b689-145e656fb9e2
trentmkelly/LessWrong-43k
LessWrong
COVID-19 estimated test detection windows for IgG, IgM and nasal PCR tests I spent a while looking for this information online and found it hard to come by, so I've compiled my research conclusions. These numbers are based on the best information I could find, but please comment with any critique or improvements. Bac...
55df7061-7f9f-4fdc-b195-13bf5a67fe97
trentmkelly/LessWrong-43k
LessWrong
Coupling for Decouplers — Intro Rationalist don't talk about dating much. At least — not on LessWrong.  This is surprising! Dating touches on every theme of interest to rationalists: psychology, game theory, epistemology, signaling, self-transformation, agency, even AI — not to mention most people's personal interest...
abd64120-5a1d-488e-94a8-77f3372259a1
trentmkelly/LessWrong-43k
LessWrong
Show LW: HackerNews but for research papers Hey there, A few days ago I went semi-viral on HN for posting a side project: https://news.ycombinator.com/item?id=40455478 Here's a screenshot of the analytics for the last few days: The site is called papertalk. Basically I made it because I'd love to have a LW/HN typ...
ff42dc73-55aa-4279-b986-720beb505ec0
trentmkelly/LessWrong-43k
LessWrong
Freedom is slavery These comparisons are sometimes made as arguments in favor of the former in each pair being forcibly prevented: * Selling equity in yourself is like slavery * Allowing organ selling is like stealing organs * Choosing genetic characteristics of your children is like eugenics * Languages dying ou...
2d86d850-0945-4e63-b60b-2eab3deff2f0
trentmkelly/LessWrong-43k
LessWrong
Meetup : Israel Less Wrong Meetup: Communities and the rationalist community Discussion article for the meetup : Israel Less Wrong Meetup: Communities and the rationalist community WHEN: 21 August 2014 07:00:00PM (+0300) WHERE: 98 Yigal Alon St., 29th floor, Tel Aviv We're going to have a meetup on Thursday, August...
935b50e1-146c-489c-9308-82eff85cfb4d
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
Are AI Safety Concerns Philosophically Sound? all right it's three o'clock so we will start so thank you everybody for joining this ai safety debate my name is aaron stuppel and i've started a web magazine called conjecture magazine that is dedicated to the philosophy of progress we think karl popper's critical ration...
072b6653-74ff-4db8-9759-29bdec2dc7e7
trentmkelly/LessWrong-43k
LessWrong
(Some) politicians might be more rational than they appear. The following excerpt from an NPR story on TARP makes me feel that while the world is mad, I might have overestimated how much of that madness comes from our political leaders: > [Neel Kashkari, the Treasury official running TARP] became a sort of punching b...
cee87cf8-db9f-4179-9081-7450338a9cce
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"The room in which Jeffreyssai received his non-beisutsukai visitors was quietly formal, impeccably appointed in only the most conservative tastes. Sunlight and outside air streamed through a grillwork of polished silver, a few sharp edges making it clear that this wall was not to be opened. The floor and walls were gl...
0446210c-63d2-43cb-9205-ce37ecd2d166
trentmkelly/LessWrong-43k
LessWrong
Age changes what you care about [Possible trigger warning for discussion of mortality.] I've been on LessWrong for a very long time.  My first exposure to the power of AI was in the mid 1980's, and I was easily at "shock level 4" at the turn of the millennium thanks to various pieces of fiction, newsgroups, and discu...
520cb9d5-f21f-4247-950a-176382584c5e
StampyAI/alignment-research-dataset/aisafety.info
AI Safety Info
What are some of the leading AI capabilities organizations? Some of the main research groups advancing AI capabilities are: - [Google DeepMind](https://www.deepmind.com/) - [OpenAI](https://openai.com/) - [Anthropic](https://www.anthropic.com/) - [Meta AI](https://ai.facebook.com/) - [Microsoft AI](https://www.mi...
b40c7fb9-7f4b-4e15-907b-8b7224a9342b
trentmkelly/LessWrong-43k
LessWrong
Wanting vs. Liking Revisited In Are Wireheads Happy? I discussed the difference between wanting something and liking something. More recently, Luke went deeper into some of the science in his post Not for the Sake of Pleasure Alone. In the comments of the original post, cousin_it asked a good question: why implement ...
0dd1b1b8-4272-488f-918f-bb6c3f20c812
trentmkelly/LessWrong-43k
LessWrong
The slingshot helps with learning The slingshot effect is a late-stage training anomaly found in various adaptive gradient optimization methods. In particular, slingshots are present with AdamW, the optimizer most widely used for modern transformer training. The original slingshot paper observes that slingshots tend t...
fe2a9395-0cd1-4b41-9ffa-0a7023f22492
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Kahneman's Planning Anecdote Today's post, Kahneman's Planning Anecdote was originally published on 17 September 2007. A summary (taken from the LW wiki):   > Nobel Laureate Daniel Kahneman recounts an incident where the inside view and the outside view of the time it would take to complete a project of ...
da9fa854-525d-47c8-af4a-fb65281f5d23
trentmkelly/LessWrong-43k
LessWrong
Another way to look at consciousness   Edit: First paragraph removed and small changes made to the rest.   I am putting forth a hypothesis is about the nature of consciousness. First I will have to tell you how I am using certain words because they are generally used in a number of ways.  'Brain' is an biological o...
b2b16920-d076-4d6c-bf05-0befdef6a49c
trentmkelly/LessWrong-43k
LessWrong
What are our outs to play to? This is an unfiltered, unsorted collection of plausible (not likely) scenarios where there are humans alive in 200 years. I have left out "We solve alignment" since it is adequately discussed elsewhere on the website. These non-alignment scenarios form a larger and larger part of the huma...
4ba4caea-5c02-49e6-85ec-b4bb2f688498
StampyAI/alignment-research-dataset/blogs
Blogs
Takeaways from safety by default interviews *By Asya Bergal, 3 April 2020* Last year, several researchers at AI Impacts (primarily Robert Long and I) interviewed prominent researchers inside and outside of the AI safety field who are relatively optimistic about advanced AI being developed safely. These interviews we...
3b7aa5b2-a85f-44e2-9b4e-655e4a9948fc
trentmkelly/LessWrong-43k
LessWrong
Help needed: nice AIs and presidential deaths A putative new idea for AI control; index here. This is a problem that developed from the "high impact from low impact" idea, but is a legitimate thought experiment in its own right (it also has connections with the "spirit of the law" idea). Suppose that, next 1st of Ap...
4489f1c7-f003-4272-97b6-988a1a92d43f
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
On Deference and Yudkowsky's AI Risk Estimates *Note: I mostly wrote this post after Eliezer Yudkowsky’s “[Death with Dignity](https://www.lesswrong.com/posts/j9Q8bRmwCgXRYAgcJ/miri-announces-new-death-with-dignity-strategy)” essay appeared on LessWrong. Since then, Jotto has written [a post](https://www.lesswrong.com...
a933436f-053a-4ecd-93e2-47aa8420a618
trentmkelly/LessWrong-43k
LessWrong
If a tree falls on Sleeping Beauty... Several months ago, we had an interesting discussion about the Sleeping Beauty problem, which runs as follows: > Sleeping Beauty volunteers to undergo the following experiment. On Sunday she is given a drug that sends her to sleep. A fair coin is then tossed just once in the cour...
a0b19804-da3d-4294-9000-cad3d70db897
StampyAI/alignment-research-dataset/aisafety.info
AI Safety Info
What is Aligned AI / Stuart Armstrong working on? One of the key problems in AI safety is that there are many ways for an AI to generalize off-distribution, so it is very likely that any arbitrary generalization will be unaligned. See the [model splintering post](https://www.alignmentforum.org/posts/k54rgSg7GcjtXnMHX/...
14822fcd-5f8b-425a-81a6-86590bcafae0
trentmkelly/LessWrong-43k
LessWrong
You are an optimizer. Act like it! In the long run, optimizers win. So act like an optimizer. Optimizers use all available resources to take optimal decisions. Optimizers are motivated to have beliefs that correspond to reality, because they are needed as inputs for the function that determines the action. If you fe...
402ce65c-8e9f-4b85-98d9-fc06cc4eff1c
trentmkelly/LessWrong-43k
LessWrong
Meetup : Washington, D.C.: Meta Meetup Discussion article for the meetup : Washington, D.C.: Meta Meetup WHEN: 26 July 2015 03:00:00PM (-0400) WHERE: National Portrait Gallery Crossposted from the mailing list: As usual, we will be congregating in the courtyard from 3:00 to 3:30 p.m., to start at 3:30. The meta m...
f7f0d2bc-305a-453b-ac18-aff2a6d8ebc6
trentmkelly/LessWrong-43k
LessWrong
How I come up with ideas I desire many things. I invent ideas to satisfy my desires. Whenever I learn something new, I compare the new fact against my database of desires. If my new fact could be used to satisfy a desire then I get an idea. Some desires are universal among humans. * I want to be rich without puttin...
ad61bfc1-e9c1-4642-8f18-f769194a2caa
trentmkelly/LessWrong-43k
LessWrong
In What Ways Have You Become Stronger? Related to: Tsuyoku Naritai! (I Want To Become Stronger), Test Your Rationality, 3 Levels of Rationality Verification. Robin and Eliezer ask about the ways to test rationality skills, for each of the many important purposes such testing might have. Depending on what's possible, ...
b4a10737-1633-4d01-b4d1-ab55ae7d4666
trentmkelly/LessWrong-43k
LessWrong
Depositions and Rationality Reading this guide on deposition strategy I was impressed by how many of the techniques lawyers use on uncooperative witnesses seem like good rationality techniques (a deposition is a sworn oral testimony taken out of court without a supervising judge). Searching LessWrong, nobody seems to...