id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
0c7a98ca-1e7c-449a-a8b2-09480e871b38 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Does GPT-4 exhibit agency when summarizing articles?
As a follow up to [A crazy hypothesis: GPT-4 already is agentic and is trying to take over the world!](https://www.lesswrong.com/posts/83cXhEy2X8zkucjcs/a-crazy-hypothesis-gpt-4-already-is-agentic-and-is-trying-to), I figured I would start experimenting.
If you re... |
2c9a2fac-43b7-4ffa-96d5-de5d69cdbae1 | trentmkelly/LessWrong-43k | LessWrong | Advice needed for Less Wrong Discussion Topic: Learning Mastery
Hi everyone,
I currently head up a Sydney Less Wrong Meetup, which is actually just me and a group of my friends trying to improve our own rationality.
Several are uni students, so I was thinking of doing powerpoint and initiating some discussion on lea... |
80cf8fbf-bdeb-467c-ad08-22ced4283900 | trentmkelly/LessWrong-43k | LessWrong | Facebook, The Rodents, and The Common Knowledge Machine
"The rationalist communities use of Facebook is an inadequate scenario. A group of people put real money into making a simple tool that could enable transitions from inadequate equilibria to better, 'adequate' ones. This tool was made in the basic sense, but had ... |
db1ba9be-f8b6-45dd-8467-5fb8e49bf7c7 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | When does rationality-as-search have nontrivial implications?
(This originated as [a comment](https://www.lesswrong.com/posts/efWfvrWLgJmbBAs3m/embedded-world-models#gBYrqXG84Fhmd7Cuh) on the post ["Embedded World-Models,"](https://www.lesswrong.com/posts/efWfvrWLgJmbBAs3m/) but it makes a broadly applicable point and... |
aab3234a-cea7-4113-8afc-66ff3d49b32d | trentmkelly/LessWrong-43k | LessWrong | Might intelligence decrease among humanity?
A common belief among members of this community seems to be that long-term, humans are likely to increase in intelligence, knowledge, and general problem-solving ability. I would like to take a moment to argue for the opposite position. This is not a strongly held belief of ... |
d03cacd9-2b14-4f6f-b399-e22ee326d02b | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "An oblong slip of newspaper had appeared between O'Brien's fingers. For perhaps five seconds it was within the angle of Winston's vision. It was a photograph, and there was no question of its identity. It was the photograph. It was another copy of the photograph of Jones, Aaronson, and Rutherford at the party function... |
202be2cb-1554-4fd3-9a90-facb9448c5e5 | trentmkelly/LessWrong-43k | LessWrong | Nate Soares on the Ultimate Newcomb's Problem
Nate Soares's distilled exposition of the Ultimate Newcomb's Problem, plus a quick analysis of how different decision theories perform, copied over from a recent email exchange (with Nate-revisions):
----------------------------------------
> You might be interested in E... |
74b89e6f-3db4-452c-ba4d-0c51f3564823 | trentmkelly/LessWrong-43k | LessWrong | Resource for the mappings between areas of math and their applications?
One of the things that I want to do while I'm in self-quarantine is to learn more math.
One thing that would be really helpful for me is a mapping between areas of math and the technical subfields that use that math.
I'd want to be able to go in... |
e2648f62-3a5c-42b9-8aad-c1716486d151 | trentmkelly/LessWrong-43k | LessWrong | Cognitive Neuroscience, Arrow's Impossibility Theorem, and Coherent Extrapolated Volition
Suppose we want to use the convergence of humanity's preferences as the utility function of a seed AI that is about to determine the future of its light cone.
We figured out how to get an AI to extract preferences from human beh... |
9447cef2-82b0-4f1d-bb12-3f258ab9e106 | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | GPT 5 Will be Released 'Incrementally' - 5 Points from Brockman Statement [plus Timelines & Safety]
yesterday Greg Brockman the president
and co-founder of openai announced the
company's ideas about releasing the
models Beyond gpt4 in the Tweet he made
lots of points of which I found five to
be particularl... |
c30c6871-33f1-425a-a28e-05ecd5813333 | trentmkelly/LessWrong-43k | LessWrong | Threat erosion
Often groups of people behave in consistent ways for a long time because they share a belief that the consistent way everyone is behaving will cause things to be bad for any individual who deviates from it.
For instance, there is a line in the sandwich shop. From a perspective so naive to our ubiquitou... |
2fc22b4e-83cd-45ac-a9d8-c1faa225416c | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Values, Valence, and Alignment
I have previously advocated for finding a mathematically precise theory for [formally approaching AI alignment](https://www.lesswrong.com/posts/7dvDgqvqqziSKweRs/formally-stating-the-ai-alignment-problem). Most recently I couched this [in terms of predictive coding](https://www.lesswrong... |
95e6c0cd-6f83-4210-b603-cffc037b8e37 | trentmkelly/LessWrong-43k | LessWrong | Can you care without feeling?
Epistemic status: Argument from behavioural analogy. I’m not claiming that AI systems feel care, only that human users interpret their behaviour as if it carries social meaning. When AI systems simulate care but fail to change behaviour in response to feedback, trust erodes. This essay re... |
203466e8-bc09-464d-acd7-070979841fc9 | trentmkelly/LessWrong-43k | LessWrong | [SEQ RERUN] Feynman Paths
Today's post, Feynman Paths was originally published on 17 April 2008. A summary (taken from the LW wiki):
> Instead of thinking that a photon takes a single straight path through space, we can regard it as taking all possible paths through space, and adding the amplitudes for every possi... |
aec01fa7-89da-4972-bd4a-0a7ea565e595 | trentmkelly/LessWrong-43k | LessWrong | Rationality Quotes September 2014
* Please post all quotes separately, so that they can be upvoted or downvoted separately. (If they are strongly related, reply to your own comments. If strongly ordered, then go ahead and post them together.)
* Do not quote yourself.
* Do not quote from Less Wrong itself, HPMoR, El... |
6d59413d-a29f-45fa-b968-7c65fe27f7df | trentmkelly/LessWrong-43k | LessWrong | Meetup : Fort Collins, Colorado Meetup Wedneday 7pm
Discussion article for the meetup : Fort Collins, Colorado Meetup Wedneday 7pm
WHEN: 02 May 2012 07:00:00PM (-0600)
WHERE: 144 North College Avenue, Fort Collins, CO 80524
Upgrading your life. Sunk costs. Dungeons and Dragons-style character builds as a learning e... |
6e446496-ea03-4610-a223-a3c5414e1a69 | trentmkelly/LessWrong-43k | LessWrong | A problem with group epistemics
(Cross-posted from my site.)
My goal here is to quickly describe a potential problem that groups of people may have when trying to collectively figure out the truth of some subject.
The basic idea is that people may use each others' beliefs to inform their own beliefs, without properl... |
8c90afbf-952e-4e6a-8abb-35de5a14da08 | trentmkelly/LessWrong-43k | LessWrong | [Link] Opacity as a defense against bias?
This article makes the claim that financial systems are opaque in order to help us overcome our risk aversion:
> Opacity is not something that can be reformed away, because it is essential to banks’ economic function of mobilizing the risk-bearing capacity of people who, if f... |
00ffd9cd-6d32-4768-a4e2-927f0e49a786 | trentmkelly/LessWrong-43k | LessWrong | Most experts believe COVID-19 was probably not a lab leak
The Global Catastrophic Risks Institute conducted an anonymous survey of relevant experts on whether they thought COVID was more likely caused by a lab accident (aka lab leak) or zoonotic spillover. Their summary, bolding is mine:
> The study’s experts overall... |
de493fe4-ff95-44c0-bf18-707d88782357 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Washington, D.C.: Slate Star Codex
Discussion article for the meetup : Washington, D.C.: Slate Star Codex
WHEN: 23 April 2017 03:30:00PM (-0400)
WHERE: Donald W. Reynolds Center for American Art and Portraiture
This week, we will be meeting in the courtyard to discuss the blog Slate Star Codex and essays t... |
638368e6-6fa1-4330-97c7-5af7b5e858bb | StampyAI/alignment-research-dataset/lesswrong | LessWrong | ChatGPT tells stories about XP-708-DQ, Eliezer, dragons, dark sorceresses, and unaligned robots becoming aligned
I’ve been spending a lot of time playing and working with ChatGPT. I am motivated, in part, by the belief that by interacting with it in systematic ways, we can gather clues about what it’s doing “under the... |
10f6bce0-1fd9-49fd-b7af-f0b747c34314 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Entanglement and intuition about words and meaning
*Cross-posted* [*from New Savanna*](https://new-savanna.blogspot.com/2023/10/entanglement-and-intuition-about-words.html)*.*
Two things have just occurred to me about my recent post, [Word meaning and entanglement in LLMs](https://new-savanna.blogspot.com/2023/10/wo... |
137a14d6-9ce1-4a05-b9f6-b2b07519c2aa | trentmkelly/LessWrong-43k | LessWrong | ChatGPT tells 20 versions of its prototypical story, with a short note on method
Cross-posted from New Savanna.
New working paper. Title above, links, abstract, and introduction below.
Academia.edu: https://www.academia.edu/108129357/ChatGPT_tells_20_versions_of_its_prototypical_story_with_a_short_note_on_method
SSR... |
cc598d91-7e35-4c9b-a1d5-793b6b638912 | trentmkelly/LessWrong-43k | LessWrong | Decision Analysis Sequence
This is the introduction (conclusion) to my decision analysis sequence. It covers (much more quickly and less completely) what you would expect to see in a semester-long course on decision making. The posts are:
1. Uncertainty: the basics of treating uncertainties as probabilities and doin... |
a3b1cc46-3ccd-4f7a-8fb1-ccca052cd4ea | trentmkelly/LessWrong-43k | LessWrong | Why is there variation in species?
This may be a question trivial for you, but: Why is there variation inside a species? On some instances, you could argue that the Nash equilibrum is mixed (for example, different men prefer different physical appearances of women, so women of different appearance can coexist), but: W... |
ff725bd3-1477-44e8-bf0e-5053a40ddb12 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Hedonic Loops and Taming RL
*Crossposted from* [*my personal blog*](https://www.beren.io/2023-04-19-Hedonic-loops-taming-RL/)*.*
Everybody knows about the hedonic treadmill. Your hedonic state adjusts to your circumstances over time and quickly reverts to a mostly stable baseline. This is true of basic physiolog... |
7eade51e-aa03-46fb-963f-a4b14cb1442a | trentmkelly/LessWrong-43k | LessWrong | [LINK] The half-life of a fact
Everything you know will, in due course, be wrong. A review of Samuel Arbesman's The Half-Life of Facts: Why Everything We Know Has an Expiration Date in Slate:
Arbesman's book expands on a piece he wrote in 2010 for the Ideas section of the Boston Globe. That short essay, called "Warni... |
5821d377-6b01-448e-8165-199783e459f7 | trentmkelly/LessWrong-43k | LessWrong | Grabby Aliens could be Good, could be Bad
Robin Hanson's Grabby Aliens is a succinct model of how and when technological life will spread. It argues that we've simply arrived too early in the universe's lifespan for other civilizations to have grown to the point of being visible to us yet, but they are there, and even... |
2bd697f2-ea59-4193-bb42-c6b662cd7d7c | StampyAI/alignment-research-dataset/arxiv | Arxiv | Efficient Querying for Cooperative Probabilistic Commitments
1 Introduction
---------------
Commitments are a proven approach to multiagent coordination (Singh [2012](#bib.bib32); Cohen and Levesque [1990](#bib.bib12); Castelfranchi [1995](#bib.bib9); Mallya and Huhns [2003](#bib.bib23); Chesani et al. [2013](#bib.... |
5b428cb3-0129-494c-92ed-89ed3a70a0cc | trentmkelly/LessWrong-43k | LessWrong | Paul's research agenda FAQ
I think Paul Christiano’s research agenda for the alignment of superintelligent AGIs presents one of the most exciting and promising approaches to AI safety. After being very confused about Paul’s agenda, chatting with others about similar confusions, and clarifying with Paul many times over... |
36837511-43ea-4c4c-b0b9-f0a4760bffe0 | trentmkelly/LessWrong-43k | LessWrong | Doublecrux is for Building Products
Previously:
* What product are you building?
2 years ago, CFAR's doublecrux technique seemed "probably good" to me, but I hadn't really stress tested it. And it was particularly hard to learn in isolation without a "real" disagreement to work on.
Meanwhile, some people seemed sk... |
1ebe742a-c216-4997-aaa3-ba1d6a555243 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Measuring hardware overhang
Measuring hardware overhang
===========================
Summary
-------
How can we measure a potential AI or hardware overhang? For the problem of chess, modern algorithms gained two orders of magnitude in compute (or ten years in time) compared to older versions. While it took the supe... |
45e5f4fb-a5cc-4f54-8110-d65fee9ec95a | trentmkelly/LessWrong-43k | LessWrong | Convincing An Illogical Being
There is a person who rejects logic as a valid way of making decisions. This person is perfectly capable of understanding logical arguments, but chooses to reject them.
Is there any way to convince this person that logic is not necessarily a bad thing?
The reason I bring this up is tha... |
0b20c41d-f6d0-45c8-acac-30ba52f4aa82 | trentmkelly/LessWrong-43k | LessWrong | Is there value in maintaining a socialization journal?
I live in a relatively small and out-of-the-way city in Canada. My geographic region only has a few dozen confirmed cases of Covid-19. Nonetheless, we are bracing for impact.
Suppose I or someone close to me becomes a confirmed case. If epidemiologist types come ... |
608a5ad7-2480-4f13-94e5-7d3ebdc65eb8 | trentmkelly/LessWrong-43k | LessWrong | Chapter 82: Taboo Tradeoffs, Final
Phoenix travel was a sensation entirely unlike Apparition or portkeys. You caught on fire - you definitely felt yourself catching on fire, even though there was no pain - and instead of burning to ashes, the fire burned all the way through you and you became fire, and then you went o... |
4999abbb-731a-4064-9c7c-1bca337e5129 | trentmkelly/LessWrong-43k | LessWrong | AI #13: Potential Algorithmic Improvements
At least two potentially important algorithmic improvements had papers out this week. Both fall under ‘this is a well-known human trick, how about we use that?’ Tree of Thought is an upgrade to Chain of Thought, doing exactly what it metaphorically sounds like it would do. In... |
4b53219a-40cb-4e87-8dae-08b65831e712 | trentmkelly/LessWrong-43k | LessWrong | Beware the science fiction bias in predictions of the future
We want to know the future, so we like reading predictions. These predictions can significantly vary in how long their time horizon is:
* Seconds to Minutes: poker players, high-frequency traders
* Minutes to Hours: day traders, in-game sports betting
* ... |
46fd9720-c2c4-40c5-a520-374fc354a94d | trentmkelly/LessWrong-43k | LessWrong | How to wash your hands precisely and thoroughly
Common instructions on handwashing are too vague for me.
> Scrub your hands for at least 20 seconds.
Sure, but how? I'm left guessing at which parts to scrub how much, possibly ruining the effect of the washing.
If you want to be precise and thorough in your handwashi... |
04b726dc-1f54-4e27-b51b-fac67ecb10e6 | trentmkelly/LessWrong-43k | LessWrong | A Quick List of Some Problems in AI Alignment As A Field
1. MIRI as central point of failure for... a few things...
For the past decade or more, if you read an article saying "AI safety is important", and you thought, "I need to donate or apply to work somewhere", MIRI was the default option. If you looked at FLI or F... |
d49efa46-bbf9-4cec-bed0-e57e819649c7 | trentmkelly/LessWrong-43k | LessWrong | TSR #9: Hard Rules
This is part of a series of posts where I call out some ideas from the latest edition of The Strategic Review (written by Sebastian Marshall), and give some prompts and questions that I think people might find useful to answer. I include a summary of the most recent edition, but it's not a replaceme... |
1cfdbd27-b519-49a2-a16d-0b915174b384 | trentmkelly/LessWrong-43k | LessWrong | Train for incorrigibility, then reverse it (Shutdown Problem Contest Submission)
The following was my submission to the AI Alignment Awards Shutdown Contest, for which it won an honorary mention prize:
My proposal is as follows:
1) pre-train a model for broad capabilities (though not to AGI)
2) create a list of di... |
40280691-030f-474b-9a75-61a47bb4c9df | trentmkelly/LessWrong-43k | LessWrong | Statistical models & the irrelevance of rare exceptions
“I don’t care about this instance - I don’t care about any instances! Life is too short to care about anything but the general case!”
Yes, the general case is drawn from instances, I’m saying that we shouldn't get caught up in the details unless they really matt... |
c48ba7f9-f20a-492c-a118-25718c6fa308 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Logical and Indexical Uncertainty
[Cross-posted](http://bywayofcontradiction.com/?p=484) on [By Way of Contradiction](http://bywayofcontradiction.com/)
Imagine I shot a photon at a half silvered mirror which reflects the photon with "probability" 1/2 and lets the photon pass through with "probability" 1/2.
Now, Im... |
b2b43618-717e-4f4c-8ffe-e0c8cb554f52 | trentmkelly/LessWrong-43k | LessWrong | Engaging Seriously with Short Timelines
It seems like transformative AI might be coming fairly soon. By transformative AI I just mean AI that will rapidly accelerate economic and technological progress. Of course, I am not ruling out a true singularity either. I am assuming such technology can be created using variant... |
fcf8df0c-bb26-4bec-b41c-4cd306cb05e3 | trentmkelly/LessWrong-43k | LessWrong | Measuring Ads Opt-Out Compliance
There's an interesting preprint out by Liu, Iqbal, and Saxena, Opted Out, Yet Tracked: Are Regulations Enough to Protect Your Privacy?. They describe an experiment they ran to figure out whether your choices on consent dialog boxes actually do anything. The idea is that if they work yo... |
47b2f1c1-0067-49c2-b7ab-bc9d882da4a3 | trentmkelly/LessWrong-43k | LessWrong | Do LLMs Implement NLP Algorithms for Better Next Token Predictions?
Do you think that base LLMs implement forms of "meta" algorithms like TF-IDF to predict the next token better?
Intuitively, it would be pretty smart to proceed that way because a text tends to be coherent in terms of style and in terms of vocabulary... |
2cefc2f1-7583-453e-8aeb-0e74a9e15c3c | trentmkelly/LessWrong-43k | LessWrong | [SEQ RERUN] Righting a Wrong Question
Today's post, Righting a Wrong Question was originally published on 09 March 2008. A summary (taken from the LW wiki):
> When you are faced with an unanswerable question - a question to which it seems impossible to even imagine an answer - there is a simple trick which can tur... |
bde00cbc-ccf8-48e9-b570-edc67fc7b908 | trentmkelly/LessWrong-43k | LessWrong | The 2020 Review
Today is the first day of the LessWrong 2020 Review. At the end of each year, we take a look at the best posts from the previous year, and reflect on which of them have stood the test of time.
As we navigate the 21st century, a key issue is that we’re pretty confused about which intellectual work is ... |
4c4dd8f2-cb16-4f68-829c-48dea548f092 | trentmkelly/LessWrong-43k | LessWrong | Why you should minimax in two-player zero-sum games
Disclaimer: I'm surely reinventing the wheel here.
Last week I wrote that I was dissatisfied with von Neumann and Morgenstern's argument for playing the minimax strategy in a two-player zero-sum game. Here's an argument I like:
We have two players A,B who play stra... |
3e87df1e-2ec9-41be-a2e1-03d2f97a1b75 | trentmkelly/LessWrong-43k | LessWrong | Morphological intelligence, superhuman empathy, and ethical arbitration
Morphological intelligence (Levin, 2022) is the ability to solve problems in the morphological space, e.g., navigate to a target morphology of the species. For example, tadpoles with scrambled craniofacial organs remodel into largely normal frogs ... |
81a03b98-dead-48e9-b4b0-4edc05f4bc09 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | RL with KL penalties is better seen as Bayesian inference
*This blog post is largely based on an [EMNLP paper](https://arxiv.org/abs/2205.11275) with Ethan Perez and Chris Buckley. It also benefited from discussions with and comments from Hady Elsahar, Germán Kruszewski, Marc Dymetman and Jérémy Scheurer.*
**TLDR**:... |
9fd71e2d-42d3-439d-a6c7-8c58dbad69ac | trentmkelly/LessWrong-43k | LessWrong | The Shard Theory Alignment Scheme
Generated as part of SERI MATS, Team Shard, under John Wentworth.
Thanks to Logan Smith, Garrett Baker, Michael Einhorn, Quintin Pope, and Thomas Kwa for chatting about related topics. All mistakes are my own.
The Alignment Scheme
The shard theory alignment scheme is:
> Master the... |
4a6e919d-ba7f-47d1-960e-5426f7171f69 | trentmkelly/LessWrong-43k | LessWrong | Navigating burnout
Burnout. Burn out? Whatever, it sucks.
Burnout is a pretty confusing thing made harder by our naive reactions being things like “just try harder” or “grit your teeth and push through”, which usually happen to be exactly the wrong things to do. Burnout also isn’t really just one thing, it’s more li... |
93d42e39-d2a7-415f-967d-4ccfd26b2d76 | trentmkelly/LessWrong-43k | LessWrong | A Naive Proposal for Constructing Interpretable AI
Epistemic Status: Extremely speculative, I'm not an experienced ML researcher
OpenAI discovered that language models can explain neurons by attaching a label. As part of their verification process, they trained simulated neurons where they used a language model to pr... |
50873f37-81aa-4406-afe7-d7e069ede721 | trentmkelly/LessWrong-43k | LessWrong | How likely is brain preservation to work?
People often ask me this. It’s a good question. It seems potentially important for anyone who doesn’t want to die, or doesn’t want other people to have to die. I don’t feel like I have a good answer, which is one reason I want to do more research. But recently I think my answe... |
1820141f-9d8f-4347-8fe6-6f48b86bef41 | trentmkelly/LessWrong-43k | LessWrong | Wanted: Foom-scared alignment research partner
Hi, my name is Mack Gallagher, I'm 20 years old, I have ZERO familiarity with computer architecture or [math or computer science technical jargon] [although I'm OK at reasoning], I have never attended a prestigious university, and I currently live in a 15,000-person meatp... |
28cb2191-c78e-41bd-b956-4a70a4dabfd9 | trentmkelly/LessWrong-43k | LessWrong | Meetup : San Francisco Meetup: Group Debugging
Discussion article for the meetup : San Francisco Meetup: Group Debugging
WHEN: 25 January 2016 06:15:00PM (-0800)
WHERE: 1597 Howard St. San Francisco, CA
We'll be meeting to try and solve each others problems! As usual, call 301-458-0764 to be let in.
This previousl... |
295ce37b-3d02-4b24-a30d-5e158a36b943 | trentmkelly/LessWrong-43k | LessWrong | Anthropic’s Recommendations to OSTP for the U.S. AI Action Plan
Note: This is an automated crosspost from Anthropic. The bot selects content from many AI safety-relevant sources. Not affiliated with the authors or their organization and not affiliated with LW.
----------------------------------------
In response to ... |
13337331-8dbc-46b0-9547-94eacccb2fa0 | StampyAI/alignment-research-dataset/blogs | Blogs | values system as test-driven development
values system as test-driven development
----------------------------------------
i realized something while reading [hands and cities on infinite ethics](https://www.lesswrong.com/posts/5iZTwGHv2tNfFmeDa/on-infinite-ethics): the work of determining the shape of [our values s... |
5c64060c-9b21-441b-bf12-9feef17ba2e4 | trentmkelly/LessWrong-43k | LessWrong | All AGI safety questions welcome (especially basic ones) [July 2022]
tl;dr: Ask questions about AGI Safety as comments on this post, including ones you might otherwise worry seem dumb!
Asking beginner-level questions can be intimidating, but everyone starts out not knowing anything. If we want more people in the worl... |
c71df191-f4f8-48be-b9a6-9d4c3b81c682 | trentmkelly/LessWrong-43k | LessWrong | Meetup: Oslo LessWrong meetup planning thread.
After some deliberation, I realized that an Oslo meetup wasn't going to happen by itself, and after my somewhat abortive (and way too late) attempt to organize a PI day HP:MoR wrapup meeting, I decided to just do it, and announce that it's happening, four weeks in advance... |
1f77217b-17be-4ffd-aba5-9f78e29c86e9 | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | 8 Ways ChatGPT 4 [Is] Better Than ChatGPT
I would not blame you if you thought
that all talk about gpt4 or chat gpt4 is
just that talk but we actually can have
a surprising amount of confidence in the
ways in which gpt4 will improve on chat
GPT by examining publicly accessible
benchmarks comparable large language
mode... |
f8932626-b440-44b4-ad0c-10dc4ad5e425 | trentmkelly/LessWrong-43k | LessWrong | Moral Philosophers as Ethical Engineers: Limits of Moral Philosophy and a Pragmatist Alternative
|
869f6acd-4215-4052-885d-b1efcbb2961e | StampyAI/alignment-research-dataset/arxiv | Arxiv | From Optimizing Engagement to Measuring Value
1. Introduction
----------------
Most recommendation engines today are based on predicting user engagement, e.g. predicting whether a user will click an item or not. However, there is potentially a large gap between engagement signals and a desired notion of *value* tha... |
4d3da260-ccd8-4390-af57-dacb7616c5be | trentmkelly/LessWrong-43k | LessWrong | Georgism Crash Course
For my first post on Less Wrong, I was advised to link to an existing blog post of mine. My Georgism Crash Course has been one of my most viewed essays. It aims to explain what Georgism is, why it would boost economic efficiency, and it addresses some of the possible drawbacks of it. All comments... |
cbd3261d-bebd-46a0-b44e-eb82f18d3eda | trentmkelly/LessWrong-43k | LessWrong | Why Real Men Wear Pink
"Fashion is a form of ugliness so intolerable we have to alter it every six months."
-- Oscar Wilde
For the past few decades, I and many other men my age have been locked in a battle with the clothing industry. I want simple, good-looking apparel that covers my nakedness and maybe even makes m... |
8d935f9c-41d4-4ae7-a717-abce924af19f | trentmkelly/LessWrong-43k | LessWrong | Control vs Selection: Civilisation is best at control, but navigating AGI requires selection
Epistemic status: Hypothesis that seemed interesting, but I am not confident about it.
In Control vs Selection, Abram Demski distinguishes between two types of optimisation. As an example of control, he mentions a heat-seekin... |
110a44c2-08d8-4951-beeb-8ba8a56cb119 | trentmkelly/LessWrong-43k | LessWrong | Bill Gates asks HS students "What are most important choices the world faces"?
http://www.impatientoptimists.org/Posts/2012/01/Do-You-Have-a-WorldChanging-Idea?WT.mc_id=1_11_2012_studentannualletter_fb&WT.tsrc=Facebook |
38d9a797-da29-4d41-9087-7191153733e6 | trentmkelly/LessWrong-43k | LessWrong | my note system
I've been told that my number of blog posts is impressive, but my personal notes are much larger than my blog, over a million words and with higher information density. Since I've had a bit of practice taking notes, I thought I'd describe the system I developed. It's more complex than some integrated so... |
2aab6970-10c2-4374-937f-ad26bb5cd15a | StampyAI/alignment-research-dataset/arxiv | Arxiv | Training verified learners with learned verifiers
1 Introduction
---------------
Neural networks are increasingly being deployed in a wide variety of applications with great success (Krizhevsky et al., [2012](#bib.bib18); Goodfellow et al., [2016](#bib.bib11)). However, recently researchers have raised concerns abo... |
d9d9b7d3-cac0-4b66-9f75-cd9ba9ae5515 | trentmkelly/LessWrong-43k | LessWrong | Solving For Meta-Ethics By Inducing From The Self
How do we determine what is good? By what method/metric? What is the argument? In this paper I will attempt to argue why and how I ground my ethics in an understanding of the self and it's values. To give a very broad and insufficient tl;dr: I'll try to create a descri... |
f1a93d02-74b0-4132-9b77-1af26495fd00 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | An implementation of modal UDT
.mjx-chtml {display: inline-block; line-height: 0; text-indent: 0; text-align: left; text-transform: none; font-style: normal; font-weight: normal; font-size: 100%; font-size-adjust: none; letter-spacing: normal; word-wrap: normal; word-spacing: normal; white-space: nowrap; float: none; ... |
dd9e8ff8-ec8a-4d25-9ef4-be07d5bbc88d | trentmkelly/LessWrong-43k | LessWrong | How does the organization "EthAGI" fit into the broader AI safety landscape?
I was surprised to have never heard of an organization working on long-term AI safety research, and can't find specific information on its work or connection to the AI safety community. |
d6ebaaa6-4fa1-4155-8653-b068eee53cbd | trentmkelly/LessWrong-43k | LessWrong | Call for discussion: Signalling and/vs. accomplishment
It seems to me that when people discover signalling, they see it everywhere and write essays about how no human activity is aimed at its stated purpose.
However, stated purposes and other sorts of useful work get done anyway, and I'm sure there are constraints wh... |
522a15d8-08cc-4b07-a6a7-3f6e4b4f128d | trentmkelly/LessWrong-43k | LessWrong | Epistemic and Instrumental Tradeoffs
Related: What Do We Mean By "Rationality?"
Epistemic rationality and instrumental rationality are both useful. However, some things may benefit one form of rationality yet detract from another. These tradeoffs are often not obvious, but can have serious consequences.
For instance... |
a204cf29-eae3-4e5a-adfe-608e92d13d7d | trentmkelly/LessWrong-43k | LessWrong | Uncompetitive programming with GPT-3
The recent DeepMind paper will presumably generate a lot of discussion, and I'm not an expert enough to completely understand the technicalities. But I still wonder how much of the breakthrough could be reduced to "tune GPT-3 for code generation instead of general text prediction".... |
d425d754-4947-4442-89ce-fe33af396f99 | trentmkelly/LessWrong-43k | LessWrong | Domains of truth
Every text about truth must begin noticing that truth is the most written-about topic in philosophy, which is due to the importance of the notion of truth.
You may or may not find the above sentence to be true. Perhaps you find some parts of the sentence to be true, but not others. Perhaps you feel i... |
4646b3ad-8307-4f2f-b475-b475fb4975ce | trentmkelly/LessWrong-43k | LessWrong | A space of proposals for building safe advanced AI
I liked Evan’s post on 11 proposals for safe AGI. However, I was a little confused about why he chose these specific proposals; it feels like we could generate many more by stitching together the different components he identifies, such as different types of amplifica... |
79452961-a2bd-490c-af14-e71edd9a5caf | trentmkelly/LessWrong-43k | LessWrong | How to build the skill and the habit of experimentation?
I want to make regular experimentation a part of my life and don't really know how. I thought that I should associate with and assist people who do run experiments (I'm interning with a psych lab and a paranormal investigator, and hope to work with some behavio... |
a9936dfe-aa8c-4374-ab21-2be374e9b1d4 | trentmkelly/LessWrong-43k | LessWrong | Relativized Definitions as a Method to Sidestep the Löbian Obstacle
The goal of this post is to outline a loophole of sorts in Gödel's Second Incompleteness (GSI) theorem (and of Löb's theorem by corollary) and give a somewhat sketchy description for how it might be exploited.
I'll begin by describing the loophole. E... |
741407a5-5d4f-49e3-a54e-ad3b821aed75 | trentmkelly/LessWrong-43k | LessWrong | An alternative way to browse LessWrong 2.0
This is something I've been tinkering with for a while, but I think it's now complete enough to be generally useful. It's an alternative frontend for LessWrong 2.0, using the GraphQL API.
Features:
* Fast, even on low-end computers and phones
* Quickly jump to new comme... |
503f9ec6-fc7d-4232-826e-69e339d2f205 | trentmkelly/LessWrong-43k | LessWrong | GPT-4o Responds to Negative Feedback
Whoops. Sorry everyone. Rolling back to a previous version.
Here’s where we are at this point, now that GPT-4o is no longer an absurd sycophant. For now.
Table of Contents
1. GPT-4o Is Was An Absurd Sycophant.
2. You May Ask Yourself, How Did I Get Here?.
3. Why Can’t We Al... |
25f3320a-6593-46dc-81da-21168dab1191 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Kidnapping and the game of Chicken
Observe the payoff matrix at right (the unit of reward? Cookies.). Each player wants to play 'A', but only so long as the two players play different moves.
Suppose that Red got to move first. There are some game... |
a39af0fe-040f-4b4a-9d30-f6ba14ab84a7 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Noticing the Taste of Lotus
Recently I started picking up French again. I remembered getting something out of [Duolingo](https://www.duolingo.com/) a few years ago, so I logged in.
Since the last time I was there, they added an “achievements” mechanic:
I noticed this by earning on... |
6d24c1f1-df1d-4d39-abbf-231fa58407fd | trentmkelly/LessWrong-43k | LessWrong | Are there easy, low cost, ways to freeze personal cell samples for future therapies? And is this a good idea?
I happened to come across this old post from 2015, that (in a sort of round about way) endorses freezing some cell samples today, because having frozen cell samples from your youth might prove important for fu... |
c790b90b-a5ee-4a0e-959c-d738addcf790 | trentmkelly/LessWrong-43k | LessWrong | HPMoE 2
HPMOE2
As usual, post solutions in ROT13. |
428e1673-cddc-40f1-a075-c42b8d175772 | trentmkelly/LessWrong-43k | LessWrong | On my AI Fable, and the importance of de re, de dicto, and de se reference for AI alignment
I wrote a fable for the EA "AI fables" contest, which raises the question of what happens when you copy values from humans to AIs, and those values contain self-referential pointers. The fable just raises the issue, and is mor... |
eb8d4498-7ba4-44d5-b9cf-2fb8af9ec557 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Sydney Rationality Dojo - October
Discussion article for the meetup : Sydney Rationality Dojo - October
WHEN: 04 October 2015 03:00:00PM (+1000)
WHERE: 10 Shepherd Street, Chippendale
Come join us for this month's Rationality Dojo, this Sunday. We've been experimenting with the format recently, to keep thi... |
409ffe32-4411-4c65-b842-76794b88fe83 | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | AI in Short: The Value Alignment Problem
meet theodore and his assistant robbie
an artificial intelligence in the shell
of a robot
theodore has a problem he loves to write
letters
but he's missing writing utensils
luckily
robbie's goal is to make him happy and
it is able to produce a limited amount
of items so theodor... |
16cce4e4-9f37-4ec7-aed6-54fdd640cb18 | trentmkelly/LessWrong-43k | LessWrong | A Straightforward Explanation of the Good Regulator Theorem
This post was written during the agent foundations fellowship with Alex Altair funded by the LTFF. Thanks to Alex, Jose, Daniel, Cole, and Einar for reading and commenting on a draft.
The Good Regulator Theorem, as published by Conant and Ashby in their 1970... |
13549b14-50d5-4d74-a0dc-c096120b1136 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Distillation of The Offense-Defense Balance of Scientific Knowledge
*Note: This is my first time trying to distil/summarize a paper. Please let me know if I've missed out on something important/misrepresented the paper.*
*This post is also available on [my blog](https://arjunyadav.net/blog/distillation-of-the-offens... |
18d400bd-0531-4fc7-800a-a609088f41fa | trentmkelly/LessWrong-43k | LessWrong | On Risk of Viral Infections from Chlorella
|
2037e01b-b3dc-46d6-b830-92934d09839b | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | A response to Matthews on AI Risk
Dylan Matthews has written [lots](http://www.vox.com/2015/8/4/9096899/cash-teach-fish) [of](http://www.vox.com/2015/6/3/8723189/john-paulson-harvard-donation) [useful](http://www.washingtonpost.com/news/wonkblog/wp/2013/05/31/join-wall-street-save-the-world/) [exploratory](http://www.... |
3082a68a-c4cc-4044-969f-ffd49149534b | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Decision Theory Breakdown—Personal Attempt at a Review
*Thanks to Vael Gates for mentoring and providing guidance on this project for the past couple months, and to Scott Garrabrant and Jack Ryan for illuminating conversations on the subject.*
I’m a student doing a research project on decision theory. This is larg... |
335db77e-2da2-40f6-a8b8-c9414fcf3a9e | trentmkelly/LessWrong-43k | LessWrong | Vote Qualifications, Not Issues
In the United States and other countries, we elect our leaders. Each individual voter chooses some criteria by which to decide who they vote for, and the aggregate result of all those criteria determines who gets to lead. The public narrative overwhelmingly supports one strategy for dec... |
0fbd3a99-ec4e-422a-965a-f3fe81592c04 | trentmkelly/LessWrong-43k | LessWrong | Are There Examples of Overhang for Other Technologies?
TL;DR: No.
What Do I Mean By ‘Overhang’?
Hardware Overhang for AI
One major concern about pausing AI development, from a purely safety perspective, is the possibility of hardware overhang.[1] Here is the concern as I understand it:
Suppose a policy were put i... |
ef9a93a9-4268-4d90-a61b-d3ad3caac938 | trentmkelly/LessWrong-43k | LessWrong | Reflection on a Consulting Workshop
Epistemic status: This is a summary / review / critique of a workshop I attended today. The workshop's materials are not publically available, so I don't expect any productive discussion regarding the workshop itself. This is mostly for myself to process the new information.
Intro... |
e9d920d3-cadb-47b3-82a7-360fafc88cdc | trentmkelly/LessWrong-43k | LessWrong | Gatekeeper Victory: AI Box Reflection
A few weeks ago, I made a post – [AI Box Experiment Are People Still Interested](https://www.lesswrong.com/posts/HrZer4yhegweXJ8RH/ai-box-experiment-are-people-still-interested) – about the AI Box Experiment. I had read about it in [Rationality: From AI to Zombies](https://www.les... |
916e6309-dd74-4fe6-80bd-31086adcb149 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Infinite Data/Compute Arguments in Alignment
*This is a reference post. It explains a fairly standard class of arguments, and is intended to be the opposite of novel; I just want a standard explanation to link to when invoking these arguments.*
When planning or problem-solving, we focus on the hard subproblems. If I’... |
7e8d65b6-c3f1-4154-86db-8812653db059 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Bratislava, Slovakia - the first LW meetup
Discussion article for the meetup : Bratislava, Slovakia - the first LW meetup
WHEN: 18 August 2012 05:00:00PM (+0200)
WHERE: Bratislava, city center
Dear Lesswrongians, we cordially invite You to the first Lesswrong meetup in Bratislava ! We just were at the Ratio... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.