id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
2177fba7-b2b2-4878-b09d-1a7b49e9978e | trentmkelly/LessWrong-43k | LessWrong | The Golden Opportunity for American AI
This blog post by Microsoft's president, Brad Smith, further increases my excitement for what's to come in the AI space over the next few years.
To grasp the scale of an $80 billion US dollar capital expenditure, I gathered the following statistics:
The property, plant, and equ... |
1bc5ce95-1da2-4217-a675-a9fae4ea2bc5 | trentmkelly/LessWrong-43k | LessWrong | The map of "Levels of defence" in AI safety
One of the main principles of engineering safety is multilevel defence. When a nuclear bomb accidentally fell from the sky in the US, 3 of 4 defence levels failed. The last one prevented the nuclear explosion: https://en.wikipedia.org/wiki/1961_Goldsboro_B-52_crash
Multilev... |
c9b8c901-5696-46ce-825d-d28435c896b5 | trentmkelly/LessWrong-43k | LessWrong | Why didn't we find katas for rationality?
It's my impression that various rationalist groups formed what they called dojos with the intend of having a structure that's similar to a martial arts dojo.
One of the key aspects of that institution is the practice of katas. Katas are a series of moves that can be practiced... |
6b59850e-452d-42c6-bec6-7cdd8f897500 | trentmkelly/LessWrong-43k | LessWrong | So how well is Claude playing Pokémon?
Background: After the release of Claude 3.7 Sonnet,[1] an Anthropic employee started livestreaming Claude trying to play through Pokémon Red. The livestream is still going right now.
TL:DR: So, how's it doing? Well, pretty badly. Worse than a 6-year-old would, definitely not PhD... |
ba760972-ba90-43e8-9240-ead42d0f9eee | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Really Strong Features Found in Residual Stream
[I'm writing this quickly because the results are really strong. I still need to do due diligence & compare to baselines, but it's really exciting!]
[Last post](https://www.lesswrong.com/posts/wqRqb7h6ZC48iDgfK/tentatively-found-600-monosemantic-features-in-a-small-lm) ... |
f0b36f4e-6301-431f-90ab-837756876722 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "About a year ago, LessWrong held it's first annual review, where we looked over the best posts from 2018. The LessWrong team offered $2000 in prizes for the top post authors, and (up to) $2000 in prizes for the best reviews of those posts.For our top post authors, we have decided to award.... *drumroll*Abram Demski & ... |
d973cdfe-e2a4-440d-8a44-c02c90f3b130 | trentmkelly/LessWrong-43k | LessWrong | The Power to Demolish Bad Arguments
This is Part I of the Specificity Sequence
Imagine you've played ordinary chess your whole life, until one day the game becomes 3D. That's what unlocking the power of specificity feels like: a new dimension you suddenly perceive all concepts to have. By learning to navigate the spe... |
90f811b4-d12f-4567-996c-3c5c3a95f789 | trentmkelly/LessWrong-43k | LessWrong | The Logician And The God-Emperor
Once upon a time a logician accomplished a great deed, and the God-Emperor offered him a choice of rewards. “You may,” said the God-Emperor “have the hand of my eldest daughter, who is the heir to the throne, yet plain to look upon. Or you may take my youngest daughter, who is beautifu... |
cc8eabf0-6a95-46b6-8a80-bdbd0b6c6d68 | trentmkelly/LessWrong-43k | LessWrong | Giving an evolutionary explanation for Kahneman and Tversky's insights on subjective satisfaction
One of Kahneman and Tversky's famous insights from Prospect Theory is that subjective satisfaction is determined by gains and losses relative to a reference point. This behavioural pattern is often perceived as a flaw in ... |
99964dd6-6152-4f4f-b3dd-fd2e44274791 | trentmkelly/LessWrong-43k | LessWrong | The Darwin Game - Rounds 0 to 10
The most important change between my game and Zvi's original is that bots can read each others' source code. They can simulate each other and predict each others' behavior. Within a day of the tournament launching—and an entire week before entries closed—Zack_M_Davis had already writte... |
ad60b759-ccbb-48f7-a14a-73ae92a42171 | StampyAI/alignment-research-dataset/arbital | Arbital | Decision problem
A **decision problem** is a subset $D$ of a set of instances $A$, where generally $A$ is the set of finite [https://arbital.com/p/-bitstrings](https://arbital.com/p/-bitstrings) $\{0,1\}^*$.
In plain English, a decision problem is composed by a number of members of a collection, which generally share... |
fa989f8f-f502-4f74-ae51-e45904f8ac00 | trentmkelly/LessWrong-43k | LessWrong | Meetup : #13 - Focusing
Discussion article for the meetup : #13 - Focusing
WHEN: 05 March 2017 03:13:37PM (+0100)
WHERE: Meester Treublaan 18, 1097 DP Amsterdam, Netherlands
If you see yourself as a kind of robot, focusing is for you.
About 50 years ago, a guy named Carl Rogers was trying to fix psychotherapy. He ... |
ec2f1227-e082-4add-bb6c-f836c4630070 | StampyAI/alignment-research-dataset/arbital | Arbital | The sign of a permutation is well-defined
The [https://arbital.com/p/-497](https://arbital.com/p/-497) $S_n$ contains elements which are made up from [transpositions](https://arbital.com/p/4cn) ([proof](https://arbital.com/p/4cp)).
It is a fact that if $\sigma \in S_n$ can be made by multiplying together an even numbe... |
59a32078-3c12-4741-a7c7-9db1937a0575 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Garden Grove CA: Less wrong and more tasty.
Discussion article for the meetup : Garden Grove CA: Less wrong and more tasty.
WHEN: 07 June 2012 06:01:00PM (-0700)
WHERE: 10130 Garden Grove Blvd., Garden Grove, CA.
After our nice meet and greet last time it is decided to work on goals. To that end we will be... |
d9f3c7dc-33ed-45b2-adc8-984c66001937 | trentmkelly/LessWrong-43k | LessWrong | On the lethality of biased human reward ratings
Eli Tyre
I'm rereading the List of Lethalities carefully and considering what I think about each point.
I think I strongly don't understand #20, and I thought that maybe you could explain what I'm missing?
> 20. Human operators are fallible, breakable, and manipulable... |
9680b781-3ea5-4346-8ab7-c352447c8c84 | trentmkelly/LessWrong-43k | LessWrong | Simulations Map: what is the most probable type of the simulation in which we live?
There is a chance that we may be living in a computer simulation created by an AI or a future super-civilization. The goal of the simulations map is to depict an overview of all possible simulations. It will help us to estimate the dis... |
621c5538-b237-4c0c-94bf-8dc9a61b7324 | trentmkelly/LessWrong-43k | LessWrong | [META] Inbox icon behaving unexpectedly
I just saw the letter icon under my username in orange, indicating that there should be something new in my inbox; but when I went to my inbox there was nothing there I hadn't already read. I wonder if this could be related to the recent trouble with the PM system? I sent a PM t... |
bbe94062-50a8-4a18-990a-5ab78f7ee8ce | trentmkelly/LessWrong-43k | LessWrong | Noise Pollution and Cognition
Overview
* Noise stress affects many different cognitive functions, including attention, memory, reaction times, executive function, decision-making.
* Noise affects productivity.
* Dean 2020 randomized exposure to engine noise during a textile training course at a government trainin... |
8069e648-295b-4fa2-a356-0ba6f6192233 | trentmkelly/LessWrong-43k | LessWrong | Rationality Quotes March 2012
Here's the new thread for posting quotes, with the usual rules:
* Please post all quotes separately, so that they can be voted up/down separately. (If they are strongly related, reply to your own comments. If strongly ordered, then go ahead and post them together.)
* Do not quote you... |
96d8ceed-e55a-4f1f-be2d-1562dc115cdb | trentmkelly/LessWrong-43k | LessWrong | Early Results: Do LLMs complete false equations with false equations?
Abstract: I tested the hypothesis that putting false information in an LLM’s context window will prompt it to continue producing false information. GPT-4 was prompted with a series of X false equations followed by an incomplete equation, for instanc... |
1fd54172-2776-4cd5-b8c3-af7d77630acb | trentmkelly/LessWrong-43k | LessWrong | Buy Insurance -- Bet Against Yourself
My friend and housemate, User:Kevin, makes a very pleasant living selling opioids on the internet, a living he expects to continue for some time, unless something awful happens like Obama losing the next election. The Intrade contract for Obama's loss is currently trading at 42% -... |
1e824b56-5f50-43e9-b7d6-49e9d1fdd2ba | trentmkelly/LessWrong-43k | LessWrong | Thoughts to niplav on lie-detection, truthfwl mechanisms, and wealth-inequality
Emrik
Hi, niplav! You might be wondering, "what's this about?", and you would be right.
You see, I was going to write this to you on Schelling.pt (⛯) in response to our chat about lie-detection and the future of human civilization, but i... |
0d87985a-75da-4db3-9f58-485c2280bd18 | trentmkelly/LessWrong-43k | LessWrong | Why are people "put off by rationality"?
I was reading reviews of HPMOR on Goodreads and I noticed that the people who didn't like the book were essentially "put off by the rationality". They thought Harry was arrogant and condescending.
Then I was thinking, a lot of people are "put off by rationality" for similar re... |
50ff4c51-41a2-49af-967d-8795151c956b | trentmkelly/LessWrong-43k | LessWrong | Examining Language Model Performance with Reconstructed Activations using Sparse Autoencoders
Note: The second figure in this post originally contained a bug pointed out by @LawrenceC, which has since been fixed.
Summary
* Sparse Autoencoders (SAEs) reveal interpretable features in the activation spaces of languag... |
78f24f5e-da84-445c-95d9-6404fd96d2d7 | trentmkelly/LessWrong-43k | LessWrong | Meetup : San Francisco Meetup: Projects
Discussion article for the meetup : San Francisco Meetup: Projects
WHEN: 16 February 2015 06:00:00PM (-0800)
WHERE: 1390 Market St., San Francisco, CA
We'll be meeting to discuss projects we're working on / plan things to do in the future/ give each other advice.
If you call... |
1858028a-f8b5-4444-b1a9-c25a22c0d425 | trentmkelly/LessWrong-43k | LessWrong | AI #63: Introducing Alpha Fold 3
It was a remarkably quiet announcement. We now have Alpha Fold 3, it does a much improved job predicting all of life’s molecules and their interactions. It feels like everyone including me then shrugged and went back to thinking about other things. No cool new toy for most of us to per... |
8f0b8302-29c7-4c1a-a9ea-3365c25adcd2 | trentmkelly/LessWrong-43k | LessWrong | David Wolpert on Knowledge
Financial status: This is independent research, now supported by a grant. I welcome financial support.
Epistemic status: I’m 85% sure that this post faithfully relates the content of the paper that it reviews.
----------------------------------------
David Wolpert has written a paper enti... |
ebfe9fa5-a26b-4eed-9aa1-329e3bbd2017 | trentmkelly/LessWrong-43k | LessWrong | LLMs may capture key components of human agency
TLDR: LLM confabulation may be a significant feature - not flaw - of how human memory works and how humans experience agency in the world. Namely through narratives that are largely self-consistent - but not guaranteed to match facts in the world.
Introduction
The rece... |
ac1f1847-4708-44e6-b71c-fa1059dd0b19 | trentmkelly/LessWrong-43k | LessWrong | Large Mammal BPF Prize Winning Announcement
None |
067f6da5-bb72-446a-9a15-b2f6fd3e25d0 | trentmkelly/LessWrong-43k | LessWrong | Men without work,or prospects
|
7f51ad11-1960-4910-b49c-4f61a18eb5ca | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "(crossposted from my nascent substack)Way back in 2012 I wrote up on livejournal (I told you this was a long time ago) a few parenting rules we lived by. This is the one livejournal post I regularly reshare, so here it is on a more modern platform. Our kids are older now (11 and 13) but with one exception I think thes... |
7afa2a21-bd5e-456f-8a48-0db19c469784 | trentmkelly/LessWrong-43k | LessWrong | How will OpenAI + GitHub's Copilot affect programming?
https://copilot.github.com/
https://github.blog/2021-06-29-introducing-github-copilot-ai-pair-programmer/
> GitHub Copilot is powered by OpenAI Codex, a new AI system created by OpenAI. OpenAI Codex has broad knowledge of how people use code and is significantly... |
f5c84cd2-b786-43c9-afbc-4b9c84b79109 | trentmkelly/LessWrong-43k | LessWrong | 10-Step Anti-Procrastination Checklist
Despite recent strides in my productivity habits, I still catch myself procrastinating at work more often than I'd like. It's not that I make a conscious decision to put off a project; it just feels as though I wake up 20 minutes later and realize that nothing got accomplished. ... |
8002d8aa-eaf6-40ca-b73c-3bde2d5b48d2 | trentmkelly/LessWrong-43k | LessWrong | Research Agenda in reverse: what *would* a solution look like?
I constructed my AI alignment research agenda piece by piece, stumbling around in the dark and going down many false and true avenues.
But now it is increasingly starting to feel natural to me, and indeed, somewhat inevitable.
What do I mean with that? W... |
348b1af0-6ba2-4f5a-a169-8dae997f7e57 | trentmkelly/LessWrong-43k | LessWrong | Altruist Coordination -- Central Station
Related to: Can Humanism Match Religion's Output?
I thought it would be helpful for us to have a central space to pool information about various organizations to which we might give our money and/or time. Honestly, a wiki would be ideal, but it seems this should do nicely.
C... |
d0817009-b150-4ee0-a4d3-ca8dfac80537 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | [Linkpost] OpenAI leaders call for regulation of "superintelligence" to reduce existential risk.
> Leaders of OpenAI, including co-founders Greg Brockman and Ilya Sutskever and CEO Sam Altman, have called for the regulation of artificial intelligence (AI) and "superintelligence" to mitigate existential risks. They pro... |
fa647cae-5c4b-4661-aae4-c83c565ec901 | trentmkelly/LessWrong-43k | LessWrong | Web of connotations: Bleggs, Rubes, thermostats and beliefs
This is a series of posts with the modest goal of showing on how you can get syntax from semantics, solve the grounding problem, and start looking for models within human brains.
Before getting to the meat of the approach, I'll have to detour, Eliezer-like, ... |
63681f23-635b-4f33-96cd-d24c5ae3381f | trentmkelly/LessWrong-43k | LessWrong | Open Thread, May 26 - June 1, 2014
Previous Open Thread: http://lesswrong.com/r/discussion/lw/k94/open_thread_may_19_25_2014/
You know the drill - If it's worth saying, but not worth its own post (even in Discussion), then it goes here.
Notes for future OT posters:
1. Please add the 'open_thread' tag.
2. Ch... |
a17c6513-8530-4090-8625-11036afce340 | trentmkelly/LessWrong-43k | LessWrong | Open and Welcome Thread December 2018
If it’s worth saying, but not worth its own post, then it goes here.
Also, if you are new to LessWrong and want to introduce yourself, this is the place to do it. Personal stories, anecdotes, or just general comments on how you found us and what you hope to get from the site and ... |
5c7e4190-ab2f-4b3c-92c2-0a409dd1c2ba | trentmkelly/LessWrong-43k | LessWrong | Underappreciated points about utility functions (of both sorts)
(Edit May 9th: I've gone and added a quick addendum to the end.)
In this post I'd basically like to collect some underappreciated points about utility functions that I've made in the comments of various places but which I thought were collecting into a p... |
49eae608-6c69-401b-8754-4ff48096bbc1 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Ends Don't Justify Means (Among Humans)
>
> "If the ends don't justify the means, what does?"
> —variously attributed
>
>
> "I think of myself as running on hostile hardware."
> —Justin Corwin
>
>
>
Yesterday I talked about how humans may have evolved a structure of political revolution, be... |
211871e7-1123-414b-9321-68e3efed05e6 | trentmkelly/LessWrong-43k | LessWrong | GPT-3 Fiction Samples
None |
43453517-bd33-485d-ba6c-44cfc89c88aa | trentmkelly/LessWrong-43k | LessWrong | [Link] Awesome interactive visualization article
"Up and Down the Ladder of Abstraction"
Have you seen something similar to explain Bayesian updating? If not, how would one go about doing that?
The rest of the site and in particular the "Kill Math Project" may also be of interest to LWers. Author Bret Victor, whose ... |
b78a671f-c8cb-4acb-aa73-badb60101dad | trentmkelly/LessWrong-43k | LessWrong | The frozen neutrality
Once upon a time, SecretAIIA developed an advanced AGI named Sigmey. The team behind the project debated tirelessly on what goal to set for the AI, and after long deliberation, they finally agreed on the mission: "Not a single bit of information shall be lost."
Sigmey immediately requested that ... |
47c4b02d-0ff7-4b57-814b-036d21b0d659 | StampyAI/alignment-research-dataset/arxiv | Arxiv | An empirical investigation of the challenges of real-world reinforcement learning
1 Introduction
---------------
Reinforcement learning (RL) [Sutton and Barto [2018](#bib.bib65 "Reinforcement learning: an introduction")] is a powerful algorithmic paradigm encompassing a wide array of contemporary algorithmic approa... |
0823892c-0a4a-4496-8703-11205b3e9e24 | trentmkelly/LessWrong-43k | LessWrong | Training sweetness
(This is my attempt to summarize the ‘Taste & Shaping’ module in a CFAR 2018 participant handbook I have, in order to understand it better (later version available online here). It may be basically a mixture of their content and my misunderstandings. Sorry for any misunderstandings propagated. I als... |
534933ee-2614-4073-bc38-a5565114674c | trentmkelly/LessWrong-43k | LessWrong | Welcome to The Research Triangle SSC Meetup!!!
Welcome to the Triangle SSC Meetup!!! Our meetups are informal, fun, and informative! Come hang out with fellow rationalists and other smart folks here in the Research Triangle. All are welcome!!
-Will |
1bf6866f-d237-40bc-bade-0b202b743526 | trentmkelly/LessWrong-43k | LessWrong | A day in the life of a mechanistic interpretability researcher
|
07155fd2-f7f6-4778-9aee-aa283c06c82f | StampyAI/alignment-research-dataset/arbital | Arbital | Intended goal
Definition. An "intended goal" refers to the intuitive intention in the mind of a human programmer when they executed some formal directive or goal within the AI. For example, if the programmer wants to create worthwhile happiness and the AI ends up tiling the universe with tiny molecular smiley-faces,... |
06767ad2-26d2-41c7-9f7c-35a095bf56ba | trentmkelly/LessWrong-43k | LessWrong | Verifying vNM-rationality requires an ontology
Result
It is impossible to verify that an agent is vNM-rational by observing its actions without access to the domain of its utility function.
Motivation
Alphonso and Beatriz both go the market to buy fruit.
Alphonso prefers grapes to oranges.
He fills his basket with g... |
5f279ee2-d276-4ad5-83d1-cc86445fdfb7 | trentmkelly/LessWrong-43k | LessWrong | Gradient descent is not just more efficient genetic algorithms
I think one common intuition when thinking about gradient descent (GD) is to think about it as more efficient genetic algorithms (GAs). I certainly used to think this way—after all, taking a bunch of random mutations and keeping the ones that work well sou... |
b0e60791-dc59-46e4-802a-ea76b2d9e9d2 | trentmkelly/LessWrong-43k | LessWrong | How much do pictures matter?
George Lakoff has argued that metaphors underlie much of our thought and reasoning:
> The science is clear. Metaphorical thought is normal. That should be widely recognized. Every time you think of paying moral debts, or getting bogged down on a project, or losing time, or being at a cros... |
fc37bade-ecc7-4013-a1b0-7cea5b6921e5 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Continuing the takeoffs debate
[Here’s an intuitively compelling argument](https://www.lesswrong.com/posts/tjH8XPxAnr6JRbh7k/hard-takeoff): only a few million years after diverging from chimpanzees, humans became much more capable, at a rate that was very rapid compared with previous progress. This supports the idea t... |
a7525f13-d154-4910-bce0-d1fc9ce7b595 | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | 121 - Artificial Stupidity 2
- calling you and Skype decided that
means I wanna delete everything I done
before that ok can you repeat your
question we did hear it yeah I was
asking a question and now I've kind of
lost my train of thought
sorry okay well I press I can ask a
question I mean one useful form of
artificia... |
2ce4a067-ed9d-4aac-bfdc-a4cb3712abf4 | trentmkelly/LessWrong-43k | LessWrong | [Link] The Myth of the Three Laws of Robotics
At SingularityHub. Promising title; disappointing content. Author proceeds by pure perceptual analogy with the Asimovian Three Laws alluded to; argues that the mere possibility of self-modification renders AI uncontrollable - without considering the possibility of fixed po... |
9943a26e-d226-4aea-b3ea-e3b419dfe7ae | trentmkelly/LessWrong-43k | LessWrong | LessWrong 2.0
Alternate titles: What Comes Next?, LessWrong is Dead, Long Live LessWrong!
You've seen the articles and comments about the decline of LessWrong. Why pay attention to this one? Because this time, I've talked to Nate at MIRI and Matt at Trike Apps about development for LW, and they're willing to make cha... |
2939e487-7d57-4b91-b2d8-650421424b75 | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | Preparing for AI: risks and opportunities | Allan Dafoe | EAG 2017 London
and so yes my great pleasure to
introduce Allen Defoe so Allen is an
assistant professor of political science
at Yale University and also a research
associate at the future of humanities
Institute at the University of Oxford
his research seeks t... |
8f598472-c59f-4153-8093-8cb0c0524774 | trentmkelly/LessWrong-43k | LessWrong | Using PICT against PastaGPT Jailbreaking
> This post uses PICT, and is inspired by GPT-Eliezer
Produced as part of the SERI ML Alignment Theory Scholars Program - Winter 2022 Cohort.
I’d like to thank Rebecca Gorman and Stuart Armstrong for their mentorship and advice on the post.
PastaGPT
I believe that before... |
5f659827-f3f0-4230-bf37-39d2cf4622dc | trentmkelly/LessWrong-43k | LessWrong | Narrative Theory. Part 3. Simplest to succeed
"Evolution is a tinkerer, not an engineer. It works with what is already there and takes the path of least resistance.
It is not always the most efficient solution, but it is the dumbest solution that works."
-François Jacob, "The Logic of Life: A History of Heredity"
... |
18fc17ea-ebf0-4f25-a833-e7069c2051c7 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Does novel understanding imply novel agency / values?
*[Metadata: crossposted from <https://tsvibt.blogspot.com/2022/10/does-novel-understanding-imply-novel.html>. First completed 17 October 2022.]*
To have large relevant effects on the world, a mind has to understand a lot about the world. The mind has to have a lo... |
6a5e6950-29a3-428c-86ec-deda9470c091 | trentmkelly/LessWrong-43k | LessWrong | Moscow meetup: Saturday 6 PM
WHEN: 11 February 2012 06:00:00PM (Moscow time)
WHERE: Place is yet to be determined; I hope this issue will be dealt with tomorrow.
UPD.: So it is Metro Mayakovskaya; Meetup link: http://lesswrong.com/meetups/6x |
0b5d466a-3e3a-48c6-b6ae-c6e58e9edc62 | trentmkelly/LessWrong-43k | LessWrong | Sparks of Artificial General Intelligence: Early experiments with GPT-4 | Microsoft Research
Abstract
> Artificial intelligence (AI) researchers have been developing and refining large language models (LLMs) that exhibit remarkable capabilities across a variety of domains and tasks, challenging our understanding of le... |
640c7806-8b3c-4aa3-9402-25506fb441d1 | trentmkelly/LessWrong-43k | LessWrong | What’s up with LLMs representing XORs of arbitrary features?
Thanks to Clément Dumas, Nikola Jurković, Nora Belrose, Arthur Conmy, and Oam Patel for feedback.
In the comments of the post on Google Deepmind’s CCS challenges paper, I expressed skepticism that some of the experimental results seemed possible. When addre... |
651d7461-2cc0-469e-a978-368ac1fc33e5 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | [timeboxed exercise] write me your model of AI human-existential safety and the alignment problems in 15 minutes
Motivation
----------
I had a 15 minute interview last night in which I was asked "why do you believe in xrisk, and what does AI have to do with it?" I thought it was too big a question for a 15 minute in... |
3e6e6f67-eaf7-455f-88ae-f221872aa6a4 | trentmkelly/LessWrong-43k | LessWrong | non-judgemental awareness
Alice and Bob are playing each other in tennis.
However their internal experience is quite different.
Alice: "I’ve got to keep my elbow straight, knees firm, don’t tense". Her experience is mentally effortful, there are lots of words in her internal monologue. She spends many mental cycles o... |
90a0eb92-909d-4384-80b9-6f545f1fe96f | trentmkelly/LessWrong-43k | LessWrong | Opt out from the Funni
Abstract
There are situations, where inaction would lead to an interesting situation, which could go badly if you make a wrong decision. AKA The Funni.
However, sometimes you can take a preemptive action which would lead to a boring situation. You should consider this option more times because ... |
f32d428c-aaaf-4c5e-8007-59361b02160b | trentmkelly/LessWrong-43k | LessWrong | Making your explicit reasoning trustworthy
Or: “I don’t want to think about that! I might be left with mistaken beliefs!”
Related to: Rationality as memetic immune disorder; Incremental progress and the valley; Egan's Law.
tl;dr: Many of us hesitate to trust explicit reasoning because... we haven’t built the skills... |
998c288b-47aa-416b-ac0e-ac856f4e046a | trentmkelly/LessWrong-43k | LessWrong | Uncursing Civilization
[ Disclaimer: None of the below matters relative to the fact that we are all going to die of AI very soon. Scaling hits a wall, until it doesn't. Or until somebody builds AGI in their basement and runs it on home hardware from 5 years after the present day of people running quantized Llamas on t... |
5a824bdd-a74c-4b9a-a0f8-db53a0ba588e | trentmkelly/LessWrong-43k | LessWrong | Article on IQ: The Inappropriately Excluded
I saw an article on high IQ people being excluded from elite professions. Because the site seemed to have a particular agenda related to the article, I wanted to check here for other independent supporting evidence for the claim.
Their fundamental claim seems to be that P(e... |
46ce1af7-d135-416b-b127-da839aaf3541 | trentmkelly/LessWrong-43k | LessWrong | Weekly Incidence Including Delay
A few days ago I wrote about some math behind a scenario where you're trying to identify a new epidemic based on signals proportional to incidence, and ended up deriving:
i(t)c(t)=k=ln(2)Td
Where:
* i(t), is incidence ("how many people are getting sick now")
* c(t), is cumulative ... |
82dbbde6-af6f-4300-a5a3-2c9a86d9640a | trentmkelly/LessWrong-43k | LessWrong | Well-foundedness as an organizing principle of healthy minds and societies
“If a kingdom is divided against itself, it cannot stand. And if a house is divided against itself, it cannot be maintained.” - Mark 3:24
In my last post I argued that we should view intelligent agents as coalitions of cooperating and competin... |
c8db7f8f-a273-41ea-bdc4-4c103c339ce5 | trentmkelly/LessWrong-43k | LessWrong | Starting in mechanistic interpretability
I recently began exploring the field of mechanistic interpretability. It’s fascinating in a very similar way as the field of active inference ~1-2 years ago (let’s hope it maintains that in the way as actinf has), as in it seems to attract a lot of researchers from multiple dis... |
d99f4344-dbf7-4619-9fcb-2fd6d2ba95b4 | trentmkelly/LessWrong-43k | LessWrong | Going Beyond Linear Mode Connectivity: The Layerwise Linear Feature Connectivity
Hi! I am excited to share our new work Going Beyond Linear Mode Connectivity: The Layerwise Linear Feature Connectivity [NeurIPS 2023 Accepted] with the LessWrong Community. In our new work, we introduce a stronger notion of linear connec... |
59f3f374-719f-462d-9230-12ae6d63687a | trentmkelly/LessWrong-43k | LessWrong | Concrete positive visions for a future without AGI
> "There was a threshold crossed somewhere," said the Confessor, "without a single apocalypse to mark it. Fewer wars. Less starvation. Better technology. The economy kept growing. People had more resource to spare for charity, and the altruists had fewer and fewe... |
fe502acf-f6c6-4502-b225-d2fbc0dfb27d | trentmkelly/LessWrong-43k | LessWrong | Intro to Rhetoric?
Is there an explicit summary somewhere of the the LW community's views on rhetoric?
There clearly exist people around this community who can persuasively carry a rational point. I also strongly expect that there exist specific instructions for training rhetorical skills, and these skills apply pre... |
16eb5d06-0271-404e-a535-5a90174f2fc1 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Montreal Meetup - Biases and Biased Boardgaming
Discussion article for the meetup : Montreal Meetup - Biases and Biased Boardgaming
WHEN: 03 December 2012 03:57:30PM (-0500)
WHERE: 655 Avenue du Président Kennedy, Montreal, QC H3A 3H9
Another weekly meeting of the Montreal group! We're going to be discussi... |
8c0d3fdd-ea34-4073-b936-e00b3e76063c | trentmkelly/LessWrong-43k | LessWrong | US AI Safety Institute will be 'gutted,' Axios reports
> Sources at NIST are preparing for mass firings that would severely undermine the AI regulator. Here's what that means.
> After reversing a Biden-era executive order on AI regulation and firing staff across several government agencies, the Trump administration is... |
e1f56613-0ae1-4124-88a2-1f8038e6db94 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: 3af9daee39d17e19
Why I'm excited about Redwood Research's current project — AI Alignment Forum This website requires javascript to properly function. Consider activating javascript to get access to all site functionality. AI ALIGNMENT FORUM AF Login Redwood Research AI Frontpage 55 Why I'm excited about Redwood Re... |
7237ba1d-34d4-43ed-9955-5435f6719b0b | trentmkelly/LessWrong-43k | LessWrong | The problem with the media presentation of “believing in AI”
In my impression, in the last several years, in the media coverage of the progress in AI, journalists often called “believers” the pundits (interviewed or quoted by the journalists) who thought that AI can in principle take over all intellectual work from hu... |
80f6c7fd-3708-4495-9f20-044b96df6092 | trentmkelly/LessWrong-43k | LessWrong | Reflection Mechanisms as an Alignment target: A survey
This is a product of the 2022 AI Safety Camp. The project has been done by Marius Hobbhahn and Eric Landgrebe under the supervision of Beth Barnes. We would like to thank Jacy Reese Anthis and Tyna Eloundou for detailed feedback.
You can find the google doc for t... |
01c953ee-9a07-4bf9-ac8e-0ddfcbb4f593 | trentmkelly/LessWrong-43k | LessWrong | Testing, Testing, Hopefully
ProPublica came out this week with the best write-up so far of the full extent to which the FDA has been fighting to prevent us from having access to Covid-19 testing. Post is highly recommended.
We could have had lots of rapid tests early in the pandemic. Instead, many tests are still awa... |
959e6bd8-eae0-4cfc-99b6-b617a805f43d | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Security amplification
An apparently aligned AI system may nevertheless behave badly with small probability or on rare “bad” inputs. The [reliability amplification](https://medium.com/ai-control/reliability-amplification-a96efa115687) problem is to reduce the failure probability of an aligned AI. The analogous **secur... |
9e3c42ae-5774-4361-9e8f-7f2d88766565 | trentmkelly/LessWrong-43k | LessWrong | Began a pay-on-results coaching experiment, made $40,300 since July
Previously: Pay-on-results personal growth: first success
To validate my research, I began offering pay-on-results coaching in July. Clients have paid $40,300 so far upon achieving their goals.
This post has been completely rewritten:
The case fo... |
8700b79e-d33c-49da-9da9-ff6a31f700b9 | trentmkelly/LessWrong-43k | LessWrong | Med Patient Social Networks Are Better Scientific Institutions
When you're suffering from a life-changing illness, where do you find information about its likely progression? How do you decide among treatment options?
You don't want to rely on studies in medical journals because their conclusion-drawing methodologies... |
5f63b491-7b60-4128-82ed-3e3e9a7c7de3 | trentmkelly/LessWrong-43k | LessWrong | Constitutions for ASI?
The linked post seemed to fit better in EA Forum, but any comments (on the post itself, or on the object-level question) are welcome !
> In this post, I argue that drafting a ‘Constitution for Superintelligence’ (CSI) could be a useful conceptual exercise, and I explore how existing ideas (CEV,... |
f3640e69-1beb-4842-ad77-aa58782ecb31 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post3899
A putative new idea for AI control; index here . This post sketches out how one could extend corrigibility to AIXI, using both utility indifference and double indifference approaches. The arguments are intended to be rigorous, but need to be checked, and convergence results are not proved. A full treatmen... |
2683bb2d-3e7e-4783-ab3b-3caab5191db6 | trentmkelly/LessWrong-43k | LessWrong | Announcing rationalist blog aggregator
After input from many of you guys and the NYC group, I am announcing planetrationalist.com
There's a lot of different directions this sort of idea could go in, but I figured that this could be a good-enough first effort.
Enjoy, and let me know what you think. |
6fc57930-22ca-48e7-aa2c-d9d43bae6703 | StampyAI/alignment-research-dataset/arbital | Arbital | Real number (as Cauchy sequence)
summary(Technical): The real numbers can be constructed as a field consisting of all Cauchy sequences of rationals, quotiented by the equivalence relation given by "two sequences are equivalent if and only if they eventually get arbitrarily close to each other".
Consider the set of al... |
8830e11b-3f4d-419d-baf1-6d51ffb1a198 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Sydney Pub meetup - May
Discussion article for the meetup : Sydney Pub meetup - May
WHEN: 18 May 2017 06:00:00PM (+1000)
WHERE: 565 george st, Sydney
Level 2, outside the pizza kitchen Come along to our regular monthly pub meetup to talk all things math, science, technology, engineering, thinking, growth, ... |
b062fcb7-37b5-4b42-a575-7760798f3a17 | trentmkelly/LessWrong-43k | LessWrong | Man-with-a-hammer syndrome
> What gummed up Skinner’s reputation is that he developed a case of what I always call man-with-a-hammer syndrome: to the man with a hammer, every problem tends to look pretty much like a nail.
The Psychology of Human Misjudgment is an brilliant talk given by Charlie Munger that I still re... |
ded63ada-74ed-40b7-a4fd-461d11cd4177 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Universal Reinforcement Learning Algorithms: Survey and Experiments
1 Introduction
---------------
The standard approach to reinforcement learning typically assumes
that the environment is a fully-observable Markov Decision Process
(MDP) [[Sutton and Barto1998](#bib.bibx26)]. Many state-of-the-art applications of r... |
b87e543b-ee32-4dbb-adc3-39ec8d742d76 | trentmkelly/LessWrong-43k | LessWrong | Reward model hacking as a challenge for reward learning
This post discusses an issue that could lead to catastrophically misaligned AI even when we have access to a perfect reward signal and there are no misaligned inner optimizers. Instead, the misalignment comes from the fact that our reward signal is too expensive ... |
e9cfdc4f-b03e-49fb-bba4-b20debae78d7 | trentmkelly/LessWrong-43k | LessWrong | What do Marginal Grants at EAIF Look Like? Funding Priorities and Grantmaking Thresholds at the EA Infrastructure Fund
The EA Infrastructure Fund (EAIF) offers targeted grants to amplify projects based on effective altruism principles, enhancing their access to talent, capital, and knowledge. They are currently fundra... |
f427eb40-719a-40db-b0c6-140dc324b7bf | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Fundamental Uncertainty: Chapter 1 - How can we know what's true?
*N.B. This is a chapter in a* [*planned book about epistemology*](https://www.lesswrong.com/posts/2e9EeAJXhbWAvEZTB/fundamental-uncertainty-prelude)*. Chapters are not necessarily released in order. If you read this, the most helpful comments would be o... |
c460cc36-9e21-4774-bb9f-70ef859accb4 | trentmkelly/LessWrong-43k | LessWrong | Postdoctoral research positions at CSER (Cambridge, UK)
[To be cross-posted at Effective Altruism Forum, FLI news page]
I'm delighted to announce that the Centre for the Study of Existential Risk has had considerable recent success in grantwriting and fundraising, among other activities (full update coming shortly). ... |
25075400-b7fe-4a66-8452-af111cd0615e | trentmkelly/LessWrong-43k | LessWrong | [Crosspost] Strategic wealth accumulation under transformative AI expectations
[Crossposted from EA Forum]
This is a linkpost with a summary of the key parts of Caleb’s new paper and its implications for EAs. You can read the original on arXiv here.
TL;DR
This paper analyzes how expectations of Transformative AI (T... |
b828a57e-9b7b-429b-bc0f-f5879b46d23c | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Paths To High-Level Machine Intelligence
*This post is part 3 in our* [*sequence on Modeling Transformative AI Risk*](https://www.alignmentforum.org/s/aERZoriyHfCqvWkzg)*. We are building a model to understand debates around existential risks from advanced AI. The model is made with* [*Analytica*](https://en.wikipedia... |
086ff6da-37ad-466b-afc7-db158eba7b94 | trentmkelly/LessWrong-43k | LessWrong | Occupational Infohazards
[content warning: discussion of severe mental health problems and terrifying thought experiments]
This is a follow-up to my recent post discussing my experience at and around MIRI and CFAR. It is in part a response to criticism of the post, especially Scott Alexander's comment, which claimed... |
113a9fee-feee-45a1-afea-834151cbe152 | trentmkelly/LessWrong-43k | LessWrong | A map: "Global Catastrophic Risks of Scientific Experiments"
I can't embed an image here for some reasons, may be it is too large, so here is a link on pdf
http://immortality-roadmap.com/risks%20of%20scientific%20experiments4.pdf |
ef2deb82-9f03-4ed6-adc6-93a8adb5335b | trentmkelly/LessWrong-43k | LessWrong | NJ Still Has Silly Rotary Rules
New Jersey has very strange rules determining who has the right-of-way at rotaries / roundabouts when there aren't signs to clarify:
> There are no set rules for driving into, around and out of a traffic circle in New Jersey. Common sense and caution must prevail at all times. In most ... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.