id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
f5709382-a554-4a7f-82d1-dfb65794b218 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Visualizing Neural networks, how to blame the bias
Background
==========
This post is strongly based on this paper, which it calls the LRP algorithm, <https://arxiv.org/pdf/1509.06321v1.pdf> [[1]](#fn9iq1bgoq7gd)
I later learned of the existence of this paper, which is even more similar to the ideas discussed here.... |
b8cd2963-98f2-4520-9bc0-d639b16ed326 | trentmkelly/LessWrong-43k | LessWrong | Russian x-risks newsletter winter 2019-2020.
As of February, 29, there were no official cases of coronavirus in Russia, except 3 cases from Diamond Princess and 2 older cases from China which recovered. This absence of the new cases seems statistically implausible given around 100 cases in Germany. There were some rum... |
9ca4373f-f336-4ab0-b01c-68fc648453d8 | trentmkelly/LessWrong-43k | LessWrong | Empirical Evidence Against "The Longest Training Run"
Summary
In this post, we empirically test Epoch AI’s theoretical model of an upper bound on AI training run lengths. According to this model, an upper bound for training run time can be estimated by assuming that the length of a training run is optimized for maximi... |
e084efaa-79ce-43e0-b450-f30dee0d12d3 | trentmkelly/LessWrong-43k | LessWrong | GPTs' ability to keep a secret is weirdly prompt-dependent
TL;DR
GPT-3 and GPT-4 understand the concept of keeping a password and can simulate (or write a story about) characters keeping the password. However, this is highly contingent on the prompt (including the characters' names, or previously asked questions). The... |
dbe883f7-de6f-4586-b904-69daf91e97e4 | trentmkelly/LessWrong-43k | LessWrong | Do I really not believe in God? Do you?
Well, I used to think that I do not believe in anything supernatural that affects what happens to me, but I'm wondering if maybe I actually do alieve in it. For example, a few days ago I had a close call in traffic, and when a collision I fully expected to happen just a second p... |
498ed361-7b4b-4df0-8043-cfde27031e83 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | From language to ethics by automated reasoning
*Posted also on the* [*EA Forum*](https://forum.effectivealtruism.org/posts/u3a6vuP9GapmKhhjR/from-language-to-ethics-by-automated-reasoning)*.*
Introduction
------------
In [Naturalism and AI alignment](https://www.alignmentforum.org/posts/Jo2LWuuGEGHHfGZCM/naturalism-... |
26197b08-37c8-402f-a6ce-56ea2d763a74 | trentmkelly/LessWrong-43k | LessWrong | Meetup : MIRI paper reading group
Discussion article for the meetup : MIRI paper reading group
WHEN: 03 April 2015 11:00:00AM (-0800)
WHERE: 109 15th Ave, Seattle Wa, 98122
https://www.facebook.com/events/654466484698076/ We will be following: https://intelligence.org/2014/12/23/new-technical-research-agenda-overvi... |
e81fd735-9ebd-4aa1-8567-44cc77da7bed | trentmkelly/LessWrong-43k | LessWrong | Meetup : Washington, D.C.: Fun & Games
Discussion article for the meetup : Washington, D.C.: Fun & Games
WHEN: 22 November 2015 03:00:00PM (-0500)
WHERE: Reynolds Center
x-posted from list. Gathering in courtyard from 3:00pm, hard start 3:30pm - until closing (7:00 pm).
We'll be hanging out, playing games, and eng... |
9df1fe6d-fe46-4110-9645-884b9ae88d06 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | GPT-4 Specs: 1 Trillion Parameters?
I know that OpenAI has tried to keep GPT-4's specs under wraps, but I've also heard some reports that anonymous sources are converging at a trillion parameters for the model. Can anyone confirm or deny, or qualify that statement?
<https://the-decoder.com/gpt-4-has-a-trillion-paramet... |
e535a95a-be49-4b19-a799-95a2c2ebbe56 | trentmkelly/LessWrong-43k | LessWrong | My current LK99 questions
So this morning I thought to myself, "Okay, now I will actually try to study the LK99 question, instead of betting based on nontechnical priors and market sentiment reckoning." (My initial entry into the affray, having been driven by people online presenting as confidently YES when the predi... |
9980f0ef-ce43-42f0-9e7b-808d44d13375 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Thoughts on safety in predictive learning
*(Many thanks to Abram Demski for an extensive discussion that helped clarify my thoughts. Thanks Abram &* [*John Maxwell*](https://www.lesswrong.com/users/john_maxwell) *for comments and criticisms on a draft.)*
Background: Why do I care?
==========================
Way back... |
74548361-8a38-45e5-8cc2-e580ae884a79 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post3829
(An idea from a recent MIRI research workshop; similar to some ideas of Eric Drexler and others. Would need further development before it's clear whether this would do anything interesting, let alone be a reliable part of a taskifying approach.) If you take an AI capable of pretty general reasoning, and y... |
f6b372b5-fd85-430e-b25a-aa93fef8d1a1 | StampyAI/alignment-research-dataset/arbital | Arbital | Stub
Stub-Class pages are tiny, often a single paragraph, or a couple of sentences. Stubs may have been created to provide a [https://arbital.com/p/-5xs](https://arbital.com/p/-5xs) for parenting, tagging, or reference purposes in a section under construction. An editor expanding a Stub page might start over from sc... |
45167591-aa8f-42b9-9777-660d9cc8a9b9 | trentmkelly/LessWrong-43k | LessWrong | Rationality Quotes July 2011
Here's the new quotes thread.
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 yourself.
* Do not quote comm... |
7d90b90d-2302-4a93-9eba-3f9882dc86fb | trentmkelly/LessWrong-43k | LessWrong | Taboo Your Words
In the game Taboo (by Hasbro), the objective is for a player to have their partner guess a word written on a card, without using that word or five additional words listed on the card. For example, you might have to get your partner to say "baseball" without using the words "sport", "bat", "hit", "pit... |
0a63b391-a87b-42f9-96e7-6b4580ff077b | trentmkelly/LessWrong-43k | LessWrong | The possible shared Craft of deliberate Lexicogenesis
[Note: crossposted from https://tsvibt.blogspot.com/2023/05/the-possible-shared-craft-of-deliberate.html.]
Words are good. Making more good words is good. Being better and faster at making more good words would be more good. Maybe we can get better and faster at ... |
2bc64580-bef4-4a1e-8195-d0991e482ff8 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Review of AI Alignment Progress
I'm having trouble keeping track of everything I've learned about AI
and AI alignment in the past year or so. I'm writing this post in part
to organize my thoughts, and to a lesser extent I'm hoping for feedback
about what important new developments I've been neglecting. I'm sure
that I... |
ccb80ce4-b6d7-4ca4-8d44-6c028c53b3cf | trentmkelly/LessWrong-43k | LessWrong | Demis Hassabis and Geoffrey Hinton Awarded Nobel Prizes
Geoffrey Hinton received the Nobel Prize in Physics for his role in creating the modern field of deep learning. This will strengthen his reputation as the "Godfather of AI" which was already used to amplify his public statements about AI risk.[1]
> The Nobel Pri... |
b85a5dd2-fe14-4d10-a2f8-404669040f19 | trentmkelly/LessWrong-43k | LessWrong | Democracy and individual liberty; decentralised prediction markets
A pair of links I found recently (via Marginal Revolution) and haven't found on LW:
http://www.cato-unbound.org/2014/03/10/mark-s-weiner/paradox-modern-individualism
https://bitcointalk.org/index.php?topic=475054.0;all
The former discusses lib... |
60c279a3-743f-4f8f-a566-55b759534376 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | TinyStories: Small Language Models That Still Speak Coherent English
Abstract
--------
> Language models (LMs) are powerful tools for natural language processing, but they often struggle to produce coherent and fluent text when they are small. Models with around 125M parameters such as GPT-Neo (small) or GPT-2 (smal... |
02ab57be-9de6-41a9-b7cb-755be0350597 | trentmkelly/LessWrong-43k | LessWrong | Median utility rather than mean?
tl;dr A median maximiser will expect to win. A mean maximiser will win in expectation. As we face repeated problems of similar magnitude, both types take on the advantage of the other. However, the median maximiser will turn down Pascal's muggings, and can say sensible things about dis... |
59f1be0d-0969-4ccc-ba8b-96790b6023f2 | StampyAI/alignment-research-dataset/arbital | Arbital | Sock-dresser search
You left your socks somewhere in your room. You think there's a 4/5 chance that they're in your dresser, so you start looking through your dresser's 8 drawers. After checking 6 drawers at random, you haven't found your socks yet. What is the probability you will find your socks in the next drawe... |
feaf2aff-989f-404d-b8b9-c8575d504fb8 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | BERI, Epoch, and FAR will explain their work & current job openings online this Sunday
*Sunday, August 21, 2 pm EST / 20:00 CET*
[*Zoom link*](https://us02web.zoom.us/j/81907654646) *|* [*Facebook event*](https://www.facebook.com/events/748180559789539/)
Join us this Sunday to learn about the work of Epoch and FAR ... |
976aa361-2394-42d7-835a-f3bcfbc25be7 | StampyAI/alignment-research-dataset/blogs | Blogs | Song Pairs that can be listened to together
Song Pairs that can be listened to together
-------------------------------------------
* From Nier: Song of the Ancients [Devola](https://www.youtube.com/watch?v=qCKEXPXtrEU) + [Popola](https://www.youtube.com/watch?v=IZdnJLdmRlI) = [both](https://www.youtube.com/watch?v=... |
8eacb77f-6f61-432b-aaa0-6203f970b1d2 | trentmkelly/LessWrong-43k | LessWrong | The Instrumental Value of Your Own Time
What is your time worth? Economists generally assume that your time is accurately valued by the market, because time can be converted into money (via wage labor) and back again (by hiring people to, e.g, do your chores for you). In this article, I argue that the economists are ... |
04844dd8-3388-453d-b13e-600188798514 | trentmkelly/LessWrong-43k | LessWrong | "Don't even think about hell"
I found an interesting article by Eliezer on arbital: https://arbital.com/p/hyperexistential_separation/
> One seemingly obvious patch to avoid disutility maximization might be to give the AGI a utility function U=V+W where W says that the absolute worst possible thing that can happen is... |
bf8e1407-f87e-415d-8b31-d3cba453e006 | trentmkelly/LessWrong-43k | LessWrong | Storable Votes with a Pay as you win mechanism: a contribution for institutional design
I joined the EA Forum in 2022, with a post describing my interests and agenda. I also declared in my first comment that in my view, among the main existential risk bottlenecks for this Dangerous Century, a critical one is instituti... |
2bf1a2ea-8e5b-41e3-af70-ba5106ede82f | trentmkelly/LessWrong-43k | LessWrong | The Eternal Grind
Previously: Eternal, and Hearthstone Economy versus Magic Economy, Categories of Sacredness, Sacred Cash
Epistemic Status: Extensive ‘research.’ About card games more than general principles. Those here for the rationality likely want to skip this.
As previously noted, the game Eternal (that is my ... |
40c828d8-0dc5-4eee-90a9-9979a638c880 | trentmkelly/LessWrong-43k | LessWrong | Two publicity ideas
Here's the easy idea-- how about a brochure about LW and SIAI? I was just at a couple of science fiction conventions, and it occurred to me that if there were a brochure, I could have printed it out and put it on the freebie table.
The hard idea is Sesame Street for rationalism-- entertaining vide... |
f484f12e-3770-4730-8a01-c8df23eac44a | trentmkelly/LessWrong-43k | LessWrong | Better Password Peppering
Epistemic status: confidant. I'm not a cryptographer, but no one raised strong objections and I thought about it a bunch since first thinking about it.
Everyone talking about password peppering is missing a better method to implement it.
whether you think peppers are good or bad, knowing t... |
c8055e4e-7a74-4f08-85de-49d0b0ff1ce2 | trentmkelly/LessWrong-43k | LessWrong | Rationality when Insulated from Evidence
Basically: How does one pursue the truth when direct engagement with evidence is infeasible?
I came to this question while discussing GMO labeling. In this case I am obviously not in a position to experiment for myself, but furthermore: I do not have the time to build up the b... |
07cec252-c06e-4f65-a148-ce3e1c178690 | trentmkelly/LessWrong-43k | LessWrong | Notes on Actually Trying
These ideas came out of a recent discussion on actually trying at Citadel, Boston's Less Wrong house.
What does "Actually Trying" mean?
Actually Trying means applying the combination of effort and optimization power needed to accomplish a difficult but feasible goal. The effort and optimizat... |
5fa811ae-8086-408b-a689-80d09903fa5f | trentmkelly/LessWrong-43k | LessWrong | [Paper]: The Technological Landscape Affecting Artificial General Intelligence
and the Importance of Nanoscale Neural Probes
I have a new paper out where I describe the technological landscape affecting AGI and implications for AI safety (largely building on Bostrom's work in Superintelligence).
Here's the abstract:
... |
6e81cd13-09ba-42c0-99a5-09d5c78f38db | trentmkelly/LessWrong-43k | LessWrong | Logical fallacies poster, a LessWrong adaptation.
Following http://lesswrong.com/lw/bwo/logical_fallacy_poster/ some people complained about
* the sarcastic illustration
* the lack of references
* the weird categorization that should rather fit a Bayesian framework
* the simplistic or even wrong definitions
* an... |
1c7e08a6-6215-4113-938d-327c110af635 | trentmkelly/LessWrong-43k | LessWrong | Covid 11/19: Don’t Do Stupid Things
There is very good news. We have a second vaccine! Both Pfizer and Moderna’s vaccines have now shown 94%+ effectiveness in treating Covid-19. Not to be outdone by Moderna’s report of 94.5% effectiveness, Pfizer has its final results and they are excellent. 95% effective, 94% effecti... |
5793f54b-3013-4df8-88f8-6ede219d97de | trentmkelly/LessWrong-43k | LessWrong | The meaning of words
This article aims to challenge the notion that the meaning of the words should and must be understood as the propositional or denotation content, in preference to the implied or connotational content. This is an assumption that I held for most of my life and which I suspect a great deal of aspirin... |
f83be6bb-a133-4c4a-a3e9-fdb9c1fd0c4c | trentmkelly/LessWrong-43k | LessWrong | Do factored sets elucidate anything about how to update everyday beliefs?
Pearl's theory of causal networks, interpreted as representing beliefs about variables connected causally, implies a precise algorithm for updating beliefs upon making an observation. Messages about how much each possibility is newly more or le... |
814bcb6e-ec57-45ce-9e7f-ab45d51c7b6c | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post1923
Summary This is a distillation post intended to summarize the article How RL Agents Behave When Their Actions Are Modified? by Eric Langlois and Tom Everitt, published at AAAI-21. The article describes Modified Action MDPs, where the environment or another agent such as a human may override the action of ... |
d4c13379-bb9d-4361-896b-d6c8f64493e1 | trentmkelly/LessWrong-43k | LessWrong | Isolated AI with no chat whatsoever
Suppose you make a super-intelligent AI and run it on a computer. The computer has NO conventional means of output (no connections to other computers, no screen, etc). Might it still be able to get out / cause harm? I'll post my ideas, and you post yours in the comments.
(This may ... |
71bc5c4d-0670-4c56-84bb-9f04f555cfae | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | AI alignment as “navigating the space of intelligent behaviour”
Tl;dr
=====
In this post, I introduce a conceptual tool for thinking about the epistemic landscape of AI alignment and then describe three epistemic strategies for making progress on the alignment problem: 1) tinkering, 2) idealisation and 3) intelligenc... |
09249879-077e-4491-ad09-91ef9481edab | trentmkelly/LessWrong-43k | LessWrong | Two Family Dance Flyers
I'm going to be calling another family dance in a week, and Lily and Anna wanted to make flyers to advertise it. I wrote out a sheet with the key details they might want to include:
Lily wanted to do hers on the computer, and it ended up being primarily about learning Inkscape:
Anna... |
3dbc0c8b-48b2-41ad-b163-350d63105d2c | trentmkelly/LessWrong-43k | LessWrong | Horizontal and Vertical Integration
TLDR: Slowing AI is easiest if AI companies are horizontally integrated, but not vertically integrated.
Some people have expressed a preference for there being only one leading AI company,[1] or for leading AI companies to become better at coordinating with one another.[2] If th... |
8d115ce4-7527-4e10-a74c-4e60c33be18d | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Part 2: AI Safety Movement Builders should help the community to optimise three factors: contributors, contributions and coordination
**Epistemic status**
Written as a non-expert to develop and get feedback on my views, rather than persuade. It will probably be somewhat incomplete and inaccurate, but it should provok... |
96601163-909a-486a-a2be-979a00a5d299 | trentmkelly/LessWrong-43k | LessWrong | Model Organisms of Misalignment: The Case for a New Pillar of Alignment Research
TL;DR: This document lays out the case for research on “model organisms of misalignment” – in vitro demonstrations of the kinds of failures that might pose existential threats – as a new and important pillar of alignment research.
If you... |
e15fbf85-bbf9-4b65-9166-4fdc08785fb2 | trentmkelly/LessWrong-43k | LessWrong | Group Rationality Diary, May 16-31
This is the public group instrumental rationality diary for May 16-31.
> It's a place to record and chat about it if you have done, or are actively doing, things like:
>
> * Established a useful new habit
> * Obtained new evidence that made you change your mind about some belief... |
bce59ad4-36b9-4e54-aa3d-0d580c965814 | StampyAI/alignment-research-dataset/blogs | Blogs | existential self-determination
existential self-determination
------------------------------
existential self-determination is a problem i have pondered about for a while ([1](core-vals-exist-selfdet.html), [2](genuineness-existselfdet-satisfaction-pick2.html)). in this post, i talk about how i've come to think abou... |
d2ec7674-1edf-44fb-ad5d-6e07625bf209 | trentmkelly/LessWrong-43k | LessWrong | The Redaction Machine
On the 3rd of October 2351 a machine flared to life. Huge energies coursed into it via cables, only to leave moments later as heat dumped unwanted into its radiators. With an enormous puff the machine unleashed sixty years of human metabolic entropy into superheated steam.
In the heart of the ma... |
65820079-4488-4aa3-a829-171b0cfc688c | trentmkelly/LessWrong-43k | LessWrong | Elderly discovered with superior memory and Alzheimer’s pathology
|
f2a542da-3003-47d6-81d2-c6748ee5ce1a | trentmkelly/LessWrong-43k | LessWrong | Threatening to do the impossible: A solution to spurious counterfactuals for functional decision theory via proof theory
EDIT: Vladimir_Nesov has commented that this idea already exists under the name of "the chicken rule". I think I'll still leave this post up because I personally like the presentation, but just know... |
65370415-e410-47a5-8e4c-6f6f90617d18 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Can we make peace with moral indeterminacy?
**The problem:**
Put humans in the ancestral environment, and they'll behave as if they like nutrition and reproducing. Put them in the modern environment, and they'll behave as if they like tasty food and good feelings. Pump heroin into their brains, and they'll behave as ... |
dcd9d9c8-718b-45fe-bc07-77627addbb21 | trentmkelly/LessWrong-43k | LessWrong | Neuroscience of human social instincts: a sketch
(If you’re in a hurry, you can just read the “Background and summary” section, and skip the other 85%.)
0. Background and summary
0.1 Background: What’s the problem and why should we care?
My primary neuroscience research goal for the past couple years has been to s... |
c1391383-bdc8-4b30-b2c7-50d06d67d87d | trentmkelly/LessWrong-43k | LessWrong | It's hard to make scheming evals look realistic for LLMs
Abstract
Claude 3.7 Sonnet easily detects when it's being evaluated for scheming on the Apollo's scheming benchmark. Some edits to evaluation scenarios do improve their realism for Claude, yet the improvements remain modest. Our findings demonstrate that truly d... |
c4875b7e-f3f8-4896-967f-b85a950b7c38 | trentmkelly/LessWrong-43k | LessWrong | Dath Ilan vs. Sid Meier's Alpha Centauri: Pareto Improvements
Epistemic status: Rambly; probably unimportant; just getting an idea that's stuck with me out there. Small dath-ilan-verse spoilers throughout, as well as spoilers for Sid Meier's Alpha Centauri (1999), in case you're meaning to get around to that.
The ide... |
7347e0e1-025c-4f69-9f26-2a65950dab99 | StampyAI/alignment-research-dataset/arbital | Arbital | Instrumental convergence
summary: Dissimilar goals can imply similar strategies. For example, whether your goal is to bake a cheesecake, or fly to Australia, you could benefit from... matter, energy, the ability to compute plans, and not dying in the next five minutes.
"Instrumentally convergent strategies" are a s... |
5acf0ada-9fd9-485c-9819-8c4b1425f6d1 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Dating Texts without Explicit Temporal Cues
1 Introduction
---------------
Temporal analysis of text has been an active area of research since the early
days of text mining with different focus in different disciplines.
In early computational linguistics research it was primarily concerned
with the fine-grained ord... |
e15159dc-bb19-408e-a41b-b17846a93624 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Imitative Reinforcement Learning as an AGI Approach
I've been thinking that reinforcement-learning-driven imitation between agents may be an explanation of human intelligence, and is worth exploring more as an approach to AGI.
It's difficult to get agents to exhibit the complex behaviors humans do with most optimiza... |
5dda1848-23ae-4898-955a-0146fa286967 | trentmkelly/LessWrong-43k | LessWrong | 1-page outline of Carlsmith's otherness and control series
Joe’s summary is here, these are my condensed takeaways in my own words. All links in this section are to the essays.
Outline
* Carlsmith tackles two linked questions:
* How should we behave towards future beings (future humans, AIs etc)?
* What shou... |
79300038-1f24-47a9-b1e5-3cc4ef963280 | trentmkelly/LessWrong-43k | LessWrong | Why is gender equality so rude?
I don’t see much anti-female sexism in my immediate surrounds; I notice more that is anti-male. But one place I have been continually put off by anti-female sexism is in attempts to promote gender equality. It seems especially prominent in efforts to seduce me to traditionally non-femin... |
574289e9-a5bd-4ee2-9487-31b20dd9d390 | trentmkelly/LessWrong-43k | LessWrong | 'Utility Indifference' (2010) by FHI researcher Stuart Armstrong
I just noticed that LessWrong has not yet linked to FHI researcher Stuart Amstrong's brief technical report, Utility Indifference (2010). It opens:
> Consider an AI that follows its own motivations. We're not entirely sure what its motivations are, but ... |
32907b86-1164-49e4-a096-23b670d21edb | trentmkelly/LessWrong-43k | LessWrong | Why Not a Natural Gas Generator?
In the comments on my posts about backup power, several people suggested looking into natural gas generators. After learning more about them, though, and about our local infrastructure, I don't think they make sense for us.
Unlike propane, which will stay liquid at room temperature wi... |
70cec545-955d-493b-83d5-8de5621d84f8 | trentmkelly/LessWrong-43k | LessWrong | Meetup : DC Meetup: NoVa, Rationality Games
Discussion article for the meetup : DC Meetup: NoVa, Rationality Games
WHEN: 30 August 2011 07:00:00PM (-0700)
WHERE: Ballston Commons Mall
Trying out a few more locations/times to better serve the DC area LW community.
We'll play rationality games for at least an hour b... |
2e9b19f1-d7c2-4618-b3c5-b7992edf3bc4 | trentmkelly/LessWrong-43k | LessWrong | LW survey: Effective Altruists and donations
> Analysis of 2013-2014 LessWrong survey results on how much more self-identified EAers donate
http://www.gwern.net/EA%20donations |
f28a85ed-5e4b-43e6-95e0-fb96acfa796d | trentmkelly/LessWrong-43k | LessWrong | Why are delicious biscuits obscure?
Crossposted from world spirit sock puppet.
I saw a picture of these biscuits (or cookies), and they looked very delicious. So much so that I took the uncharacteristic step of actually making them. They were indeed among the most delicious biscuits of which I am aware. And yet I don... |
b40d9c41-c028-44c6-b850-4213c9034ec4 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | The case for removing alignment and ML research from the training dataset
*This is crossposted from* [*my personal blog*](https://www.beren.io/2023-05-28-The-case-for-removing-AI-alignment-and-ML-from-the-training-dataset/)*.*
Many possible sources of danger from AI stem from the AI knowing a lot about itself, us, an... |
0e6faa18-8677-4aad-aa03-5338f702a54b | trentmkelly/LessWrong-43k | LessWrong | SIA > SSA, part 1: Learning from the fact that you exist
(Cross-posted from Hands and Cities)
This post is the first in a four-part sequence explaining why I think that one prominent approach to anthropic reasoning is better than another. Consider:
> God’s extreme coin toss: You wake up alone in a white room. There’... |
ae89e679-f517-4404-8063-115a4aaa3a43 | trentmkelly/LessWrong-43k | LessWrong | What is ontology?
Over the years I've picked up on more and more phrases that people on LessWrong use. However, "ontology" is one of them that I can't seem to figure out. It seems super abstract and doesn't seem to have a reference post.
So then, please ELI5: what is ontology? |
d0f90d0c-43df-46ef-b61c-35a91ffc68a9 | trentmkelly/LessWrong-43k | LessWrong | Notice your everything
[epistemic status: mostly me noticing (ha) a common pattern among advice i've heard and my personal experience.]
There is one skill or category of skill that is probably worth more than every other general skill of instrumental rationality on this website combined: Noticing things.
You probabl... |
b1dba9c0-bff4-490d-9fad-931b4482cd38 | trentmkelly/LessWrong-43k | LessWrong | GPT-4o Can In Some Cases Solve Moderately Complicated Captchas
Here are several examples; I found these captchas via the web rather than generating them anew, but none of them came attached to solutions so I'm not sure their presence in the training data would affect things in any case. (That said, it's possible that ... |
1b20c8ab-5de6-4e5e-b751-326c2c7a4e30 | trentmkelly/LessWrong-43k | LessWrong | Weekly LW Meetups: Berkeley, Hyderabad, Madison, New Jersey, Salt Lake City, Tucson, Washington DC
This summary was posted to LW main on July 6th, and has now been moved to discussion.
There are upcoming irregularly scheduled Less Wrong meetups in:
* Hyderabad Meetup: 08 July 2012 03:00PM
* Washington DC Meetup-Po... |
1be18adb-8b57-4bb8-b3ae-1855f270e9ab | trentmkelly/LessWrong-43k | LessWrong | My experience at ML4Good AI Safety Bootcamp
TLDR
* Experience was fantastic: I learnt a lot, made new friends and potential collaborators, and it was the step-change I needed to kickstart my contributions to AI safety.
* Big lesson for me is that mechanistic interpretability (mech interp) is not the only way technic... |
c03c0e35-ab89-401f-ba6e-2a9c1ce2be6b | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Apply to the Constellation Visiting Researcher Program and Astra Fellowship, in Berkeley this Winter
> *This is a link post for two AI safety programs we’ve just opened applications for:* [*https://www.constellation.org/programs/astra-fellowship*](https://www.constellation.org/programs/astra-fellowship) *and* [*https:... |
e27a5782-8b5e-4bb7-9d3e-ec591bc19970 | trentmkelly/LessWrong-43k | LessWrong | Recent neuroscience relevant to metaethics and extrapolation algorithms
Some Yudkowskian metaethicists have supposed that the brain does not compute value for actions or states of affairs at all, but instead uses purely behavioristic methods (via reward signals) to motivate action, and thus the 'terminal goals' we cou... |
08b0d959-188b-4f4d-b13a-7a0d80d71dde | trentmkelly/LessWrong-43k | LessWrong | Inaccessible information
Suppose that I have a great model for predicting “what will Alice say next?”
I can evaluate and train this model by checking its predictions against reality, but there may be many facts this model “knows” that I can’t easily access.
For example, the model might have a detailed representation... |
0e8c599d-47bb-4764-9b61-1b2f4b3afc58 | trentmkelly/LessWrong-43k | LessWrong | [LINK] The Selected Papers Network
John Baez has been writing, here and here, about problems with the academic journal system and a tool that might be a step towards fixing them:
> Last time Christopher Lee and I described some problems with scholarly publishing. The big problems are expensive journals and ineffectiv... |
a4faeeff-3eed-4966-8714-b3ff68608b99 | trentmkelly/LessWrong-43k | LessWrong | The impact of you is determined by them.
The title is partially not true because how you act will definitely show your image and people will judge it with common sense, but the other part is only up to them to decide the impact of your actions which makes it already terrifying in a way of not knowing what something re... |
f08a444d-5e31-4ff8-a348-8a2108ba0348 | trentmkelly/LessWrong-43k | LessWrong | [AN #149]: The newsletter's editorial policy
Alignment Newsletter is a weekly publication with recent content relevant to AI alignment around the world. Find all Alignment Newsletter resources here. In particular, you can look through this spreadsheet of all summaries that have ever been in the newsletter.
Audio vers... |
a7569da6-2d15-4e90-b700-d49501f51737 | trentmkelly/LessWrong-43k | LessWrong | Confabulation
When someone asks me why I did or said something I usually lie because the truthful answer is "I don't know". I literally don't know why I make >99% of my decisions. I think through none of these decisions rationally. It's usually some mixture of gut instinct, intuition, cultural norms, common sense and ... |
1e33bd6a-a4c5-43eb-896b-173df33b703e | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post13
This post presents a mildly edited form of a new paper by UK AISI's alignment team (the abstract, introduction and related work section are replaced with an executive summary). Read the full paper here. Executive summary AI safety via debate is a promising method for solving part of the alignment problem fo... |
42561a63-f3e1-419e-8f02-df46b9959127 | trentmkelly/LessWrong-43k | LessWrong | Don't Condition on no Catastrophes
,,,,
I often hear people say things like "By what date do you assign 50% chance to reaching AGI, conditioned on no other form of civilizational collapse happening first?" The purpose of this post is to make this question make you cringe.
I think that most people mentally replace the... |
c0fb8b4c-7ce2-43e5-8752-2ebfdd0d9f2a | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | How DeepMind's Generally Capable Agents Were Trained
Intro
=====
One of DeepMind's latest papers, [Open-Ended Learning Leads to Generally Capable Agents](https://deepmind.com/research/publications/2021/open-ended-learning-leads-to-generally-capable-agents), explains how DeepMind produced agents that can successfully... |
3eb093b3-2e18-427f-b677-af96da99d60c | trentmkelly/LessWrong-43k | LessWrong | Ruling Out Everything Else
Clear communication is difficult. Most people, including many of those with thoughts genuinely worth sharing, are not especially good at it.
I am only sometimes good at it, but a major piece of what makes me sometimes good at it is described below in concrete and straightforward terms.
Th... |
ce90fe04-7680-4fb3-bbac-ef75cf008497 | trentmkelly/LessWrong-43k | LessWrong | This is your brain on ambiguity
Let's look at one more optical illusion which reveals important features of how our brains perform inference, and suggests how better awareness of these processes of inference can lead to improved thinking, including in our daily lives. This time around the theme is ambiguity.
The spi... |
644aca8d-b547-4df1-8b9d-cb26d44c4b88 | trentmkelly/LessWrong-43k | LessWrong | Three Stories about not Thinking too Hard
I.
5 pirates split a treasure of 100 gold coins. The crew splits their booty according to sacred pirate tradition: The oldest pirate proposes how to share the coins, and ALL pirates (including the oldest) vote for or against it. If at least half vote for it, then the coins wi... |
68a1d228-0f32-4d07-ad12-63f81075a775 | trentmkelly/LessWrong-43k | LessWrong | Eternal, and Hearthstone Economy versus Magic Economy
The game Eternal (that is my referral link), created by Magic professionals including lead designer Patrick Chapin, is modern Magic: The Gathering, with some simplifications and tweaks, on a phone, with a Hearthstone interface and economy.
That is super high prais... |
331a461e-a4a4-4c82-88fd-8bfd52da708f | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | the scaling “inconsistency”: openAI’s new insight
I’ve now read the new [OpenAI scaling laws paper](https://arxiv.org/pdf/2010.14701.pdf). Also, yesterday I attended a fun and informative lecture/discussion with one of the authors.
While the topic is on my mind, I should probably jot down some of my thoughts.
This ... |
abe2e6f7-01c1-49ae-b48f-b6081a0eb970 | trentmkelly/LessWrong-43k | LessWrong | Why do we have the NATO logo?
Why does LessWrong use that little compass thingy that serves as the North Atlantic Treaty Organization's logo? (except with four additional spikes added diagonally) Was it just a coincidence? |
2fd4f016-e098-4605-96d6-44697879ce20 | trentmkelly/LessWrong-43k | LessWrong | Hunch seeds: Info bio
Just some stuff I found online. Not sure that all of these are actually good, but they seem promising as hunch inspiration. As usual, I'm sharing this because I am not going to be able to be the brain who uses these things to come to non-crackpot conclusions, but people who don't have enough hunc... |
ebb9b958-ecbf-4d90-b2c1-40d1bbf6919c | trentmkelly/LessWrong-43k | LessWrong | The institution of email
There seems to be a common phenomenon where people get messages, then fail to respond to them, then feel bad. And the rarer strategy of actually dealing with all of one’s emails promptly doesn’t even seem obviously better. Was that how things were with letters or telegrams? Is it just that the... |
e4451bfe-0c43-4d32-a6f8-8f8bda57da44 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Fundamental question: What determines a mind's effects?
*[Metadata: crossposted from <https://tsvibt.blogspot.com/2023/04/fundamental-question-what-determines.html>. First completed April 9, 2023.]*
A mind has some effects on the world. What determines which effects a mind has? To eventually create minds that have l... |
cf804b83-a963-438c-ba0b-027bca2979a1 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Who owns OpenAI's new language model?
OpenAI published about [their very general language-writing & comprehension model](https://blog.openai.com/better-language-models/) ([a](https://web.archive.org/web/20190214171619/https://blog.openai.com/better-language-models/)) today.
Interestingly, they're not releasing the mo... |
a1dbe352-7465-42f6-abe8-2651ec634a30 | trentmkelly/LessWrong-43k | LessWrong | Will we run out of ML data? Evidence from projecting dataset size trends
Summary: Based on our previous analysis of trends in dataset size, we project the growth of dataset size in the language and vision domains. We explore the limits of this trend by estimating the total stock of available unlabeled data over the ne... |
cb0c3f7f-ee47-4230-8ae3-5f064a8f57a7 | trentmkelly/LessWrong-43k | LessWrong | Saturating the Difficulty Levels of Alignment
TL;DR
We don't know exactly how hard alignment is, therefore it seems good to have people work on different solutions, as one axis in which solutions can differ is in, what level of difficulty they break. Some solutions would work in worlds where alignment is easy. Others... |
50dacd70-6c65-4959-a9d7-50f9649d89be | trentmkelly/LessWrong-43k | LessWrong | [Concept Dependency] Edge Regular Lattice Graph
This is a Concept Dependency Post. It may not be worth reading on its own, out of context. See the backlinks at the bottom to see which posts use this concept.
----------------------------------------
Also known as periodically labeled lattice graphs in graph theory.
... |
f1d07938-358e-41f7-9b2c-0b589d74bced | trentmkelly/LessWrong-43k | LessWrong | Robin Hanson's "Overcoming Bias" posts as an e-book.
At Luke Muehlhauser's request, I wrote a script to scrape all of Robin Hanson's posts to Overcoming Bias into an e-book; here's a first beta release. Please comment here with any problems—posts in the wrong order, broken links, bad formatting, missing posts. Thanks... |
4d4b4595-7655-4888-9029-3cda76a77a80 | trentmkelly/LessWrong-43k | LessWrong | [link] Join Wall Street. Save the World
A very interesting article on "earning to give", featuring LessWrong members Jeff Kaufman, Julia Wise, Holden Karnofsky, William MacAskill and Toby Ord. Some excerpts:
> Jason Trigg went into finance because he is after money — as much as he can earn.
>
> The 25-year-old cer... |
f247770e-4490-41b5-8ee4-2bdb888d5100 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | ELK First Round Contest Winners
Thank you to all those who have submitted proposals to the ELK proposal competition. We have evaluated all proposals submitted before January 14th[[1]](#fn-5CZvwshYeWEKxSPpz-1). Decisions are still being made on proposals submitted after January 14th.
The deadline for submissions is F... |
d267f473-cca3-44d2-9e7a-58a3a51025db | StampyAI/alignment-research-dataset/arxiv | Arxiv | Applying Deep Learning To Airbnb Search
1. Introduction
----------------
The home sharing platform at Airbnb is a two sided marketplace for hosts to rent out their spaces, referred to as listings, to be booked by prospective guests from all around the world. A typical booking starts with the guest issuing a search ... |
ebd0b0f6-55b8-4ca6-b023-0c08e1766bb7 | trentmkelly/LessWrong-43k | LessWrong | Wikidebate
|
60d11227-a4d2-4888-86c3-30a0530eb88a | trentmkelly/LessWrong-43k | LessWrong | Mildly Photochromic Lenses?
After wearing glasses for about half my life, I decided to try photochromic lenses. These darken when exposed to sunlight, as if I had put on sunglasses. It's pretty neat!
I got them in early summer, and I have mixed feelings. It was nice to be able to hang out in really bright sunlight ... |
7e4bef8f-c9d0-46cd-af51-74d50986732d | trentmkelly/LessWrong-43k | LessWrong | bridgeplan - a writeup on things one can actually do to increase lifespan
http://www.bridgeplan.org/?page_id=42 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.