id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
89039953-0511-4ea2-b84a-6a3d268ac759
trentmkelly/LessWrong-43k
LessWrong
Are the LLM "intelligence" tests publicly available for humans to take? I've seen a lot of news lately about the ways that particular LLMs score on particular tests. Which if any of those tests can I go take online to see how my performance on them compares to the models?
7e4034c3-e419-41e8-99ef-346cc058dde5
trentmkelly/LessWrong-43k
LessWrong
[Link] More Right launched Various people (including Konkvistador who has been talking about it the most) have launched their blog More Right "A group blog, More Right is a place to discuss the many things that are touched by politics that we prefer wouldn’t be, as well as right wing ideas in general. It grew out of ...
c0ee8ee8-3a32-4502-b305-c9185a036a84
trentmkelly/LessWrong-43k
LessWrong
Uncontrollable AI as an Existential Risk The following text is meant to be an introduction to existential risks from advanced AI for professionals outside of the AI safety community, in particular scientists in other relevant fields, politicians, and other decision-makers. The aim is to raise awareness of the problem ...
b0a8878d-a41f-406c-87fb-98a75fe87aaf
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
DeepMind team on specification gaming [Specification gaming: the flip side of AI ingenuity](https://deepmind.com/blog/article/Specification-gaming-the-flip-side-of-AI-ingenuity)
f0a44d1b-2719-4869-8ab4-febac4b5c2dc
trentmkelly/LessWrong-43k
LessWrong
When is further research needed? Here's a simple theorem in utility theory that I haven't seen anywhere. Maybe it's standard knowledge, or maybe not. TL,DR: More information is never a bad thing. The theorem proved below says that before you make an observation, you cannot expect it to decrease your utility, but you...
70aa4bb8-ed0e-4299-ac79-10b6f4b41bd7
trentmkelly/LessWrong-43k
LessWrong
Open Thread, August 2010 This thread is for the discussion of Less Wrong topics that have not appeared in recent posts. If a discussion gets unwieldy, celebrate by turning it into a top-level post.
95c62750-cac2-4be3-b2b2-be90df46920e
trentmkelly/LessWrong-43k
LessWrong
Revisiting shared experiences In crafting my original world modelling of love, there was a blindspot due to the significant age of my relationships. Looking back in the long run, the power of shared experiences looks like a log curve, but the reality is that different experiences have different power and it only looks...
3f01aaf3-81bc-4bb1-bde4-b9a61ee83dfd
trentmkelly/LessWrong-43k
LessWrong
List of Problems That Motivated UDT I noticed that recently I wrote several comments of the form "UDT can be seen as a step towards solving X" and thought it might be a good idea to list in one place all of the problems that helped motivate UDT1 (not including problems that came up subsequent to that post).  * decis...
d82acb35-163c-4ebe-b442-abd2b97c4eee
trentmkelly/LessWrong-43k
LessWrong
Metarationality: a messy introduction This post tries to address 'all we got were these lousy Kegan levels' critiques of metarationality by pointing at a bunch of interesting concrete topics and shouting excitedly. It's pretty long, tl;dr below: * How we think we think, vs. how we actually think: if you look closel...
12895a6f-60d0-4e5a-b569-1235ca996e33
trentmkelly/LessWrong-43k
LessWrong
How to become an AI safety researcher Crossposted from the EA Forum. Reminder that you can listen to this post and other LessWrong posts on your podcast player with The Nonlinear Library. What skills do you need to work on AI safety? And what can we learn from the paths people have taken into the field? We were insp...
0261b35a-dfde-418c-a6a9-67fc3085db66
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Moore's Paradox Today's post, Moore's Paradox was originally published on 08 March 2009. A summary (taken from the LW wiki):   > People often mistake reasons for endorsing a proposition for reasons to believe that proposition. Discuss the post here (rather than in the comments to the original post). T...
975b56e0-afa2-4937-814e-7019f091c10b
trentmkelly/LessWrong-43k
LessWrong
Principles of Disagreement Followup to:  The Rhythm of Disagreement At the age of 15, a year before I knew what a "Singularity" was, I had learned about evolutionary psychology.  Even from that beginning, it was apparent to me that people talked about "disagreement" as a matter of tribal status, processing it with th...
76ebbe8e-d335-4fa2-b5c5-fd1a0a7702c9
trentmkelly/LessWrong-43k
LessWrong
Thoughts on the Double Impact Project How it works You donate to the Republican or Democratic pool. The pool with the highest amount donates what it has more at the margin to their political party. Everything else goes to other charities. For a better and more complete explanation, check out: https://doubleimpact.char...
18857322-bff9-492d-aab7-9dae02f4b690
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"(cross posted on my roam blog)There's an old Paul Graham Essay, "Keep Your Identity Small". It's short so it's worth it to read the whole thing right now if you've never seen it. The yisbifiefyb ("yeah it's short but i'm functionally illiterate except for your blog") is roughly "When something becomes part of your ide...
a3d5c292-7215-4736-8893-f6fe0cdb97f7
trentmkelly/LessWrong-43k
LessWrong
Notification update and PM fixes Hey everyone, I just pushed an update that changes the layout of the notifications section to something that should be significantly easier to navigate (and after we've seen it work well for a few days, we will also add a small number next to the bell-icon that tells you how many unre...
426a05cb-468b-4b2b-9863-a27134ed73be
trentmkelly/LessWrong-43k
LessWrong
What's going on here? Edit: Please stop upvoting me. I'm beyond where I was when this whole mess started. Thanks, and I feel really stupid for the way I felt when I wrote this first paragraph now.   First of all, I'm sorry for this thread. I largely expect it to get downvoted, but I am considering leaving LW over th...
1495c3ae-ac3c-4b6e-8e3b-f3e3e898bbd7
StampyAI/alignment-research-dataset/lesswrong
LessWrong
GPT-3 Gems GPT-3 is captivating, and not just because of its [potential reasoning abilities](https://www.lesswrong.com/posts/L5JSMZQvkBAx9MD5A/to-what-extent-is-gpt-3-capable-of-reasoning). This post will be a living collection of my favorite experiences with the network.  Bold text is my input; square brackets conta...
61c57ce7-627b-48c9-9505-0319bacd0074
StampyAI/alignment-research-dataset/arbital
Arbital
Needs parent This page is not attached to an appropriate parent page.
c326ae5a-430e-45cb-8e3b-435ec4043343
trentmkelly/LessWrong-43k
LessWrong
Automation collapse Summary: If we validate automated alignment research through empirical testing, the safety assurance work will still need to be done by humans, and will be similar to that needed for human-written alignment algorithms. Three levels of automated AI safety Automating AI safety means developing some...
541cb0eb-d255-435e-a2a3-ce391af90b50
trentmkelly/LessWrong-43k
LessWrong
On Defense Mechanisms There is much talk about cognitive heuristics and biases in the Rationality community, whereas psychodynamic/psychoanalytic perspectives of psychology tend to be dismissed as "bad science". Now, I will agree that some Freudian ideas like penis envy and Oedipus complex are silly -- but that doesn'...
56a27882-5f78-4acf-98c2-c87ed60b31c5
trentmkelly/LessWrong-43k
LessWrong
Be more effective by learning important practical knowledge using flashcards I made a collection of 200+ flashcards (Part 1 together with part 2) to help remember useful life lessons from hundreds of great books, videos, podcasts and other sources. It includes knowledge to * Make more informed decisions. * Determi...
26d2cfd8-56b9-441e-ac5b-5868a90ba0e4
StampyAI/alignment-research-dataset/lesswrong
LessWrong
A few misconceptions surrounding Roko's basilisk There's a new [LWW page](http://wiki.lesswrong.com/wiki/Roko's%20basilisk) on the Roko's basilisk thought experiment, discussing both Roko's original post and the fallout that came out of Eliezer Yudkowsky banning the topic on *Less Wrong* discussion threads. The wiki p...
4fec0488-1b2e-49a3-99c6-d9d642e23df1
trentmkelly/LessWrong-43k
LessWrong
On desiring subjective states (post 3 of 3) Carol puts her left hand in a bucket of hot water, and lets it acclimate for a few minutes.  Meanwhile her right hand is acclimating to a bucket of ice water.  Then she plunges both hands into a bucket of lukewarm water.  The lukewarm water feels very different to her two ha...
399ebd5f-f831-43c4-b26d-eb227ea1ad8b
StampyAI/alignment-research-dataset/arxiv
Arxiv
Auditing Visualizations: Transparency Methods Struggle to Detect Anomalous Behavior 1 Introduction --------------- Neural networks with similar validation accuracy can behave very differently when deployed. A network can have a high validation accuracy, but be vulnerable to distribution shift, backdoors, or spuriou...
080ba4e4-1dff-4fb3-93d3-83f2374e3e07
trentmkelly/LessWrong-43k
LessWrong
Wealth from Self-Replicating Robots I have high confidence that economically-valuable self-replicating robots are possible with existing technology: initially, something similar in size and complexity to a RepRap, but able to assemble a copy of itself from parts ordered online with zero human interaction. This is impo...
a304e645-1952-4830-92d0-6c7213df4240
trentmkelly/LessWrong-43k
LessWrong
[Prediction] The Consequences of Radical Reform Scott Alexander suggests readers make a prediction. > Under Napoleon, the revolutionary French took over large swathes of Europe. They abolished their client states' traditional systems, replacing them with the Napoleonic Code and other "modern" legal systems. Europe at...
c09be4f3-02be-40d4-b322-f2ad05a85bfe
trentmkelly/LessWrong-43k
LessWrong
2020 LessWrong Demographics Survey Results Thank you to everyone who participated! You can see the original post here, but the survey is now closed. Here are the results: Google forms results: https://docs.google.com/forms/d/e/1FAIpQLSfHJtg9ju6cCDl7T1PZvuBC9aIYIEN4Qb_MoQFPa9iMkUIFLQ/viewanalytics Copy of the origina...
800fd086-5147-43ec-8ed4-7eeb2b88f219
trentmkelly/LessWrong-43k
LessWrong
Confusion about Sequences and Review Sequences I just made a series of attempted actions on the site that ended up with me being confused. Am I missing ways to use the site, or are these ways missing? I was reading the page for the 2019 review book launch, and someone commented that the table of contents has no links...
4f820eba-49b9-4416-9edc-4b5a7dc9f56c
trentmkelly/LessWrong-43k
LessWrong
Your Preferences Hello, I am curious about your most basic preferences. I would like you to state them as many as possible.  You can use relation signs (<, >, ...), E. g.: true knowledge > ignorance > false knowledge, which would mean that you prefer true knowledge over ignorance, and ignorance over false knowledge.
315afc59-6009-4f22-ac12-e285f0120494
trentmkelly/LessWrong-43k
LessWrong
Occam's Razor The more complex an explanation is, the more evidence you need just to find it in belief-space. (In Traditional Rationality this is often phrased misleadingly, as “The more complex a proposition is, the more evidence is required to argue for it.”) How can we measure the complexity of an explanation? How ...
0948a0ee-ad30-477e-92db-653ac2fbea2a
trentmkelly/LessWrong-43k
LessWrong
Declare your signaling and hidden agendas Follow-up to: It's okay to be (at least a little) irrational Many science journals require their authors to declare any competing interests they happen to have. For instance, if you're submitting a study about the health effects of tobacco, and you happen to sit on the boar...
f6e6a6b4-73e4-4d39-923e-1e19a4e4283c
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
Eliezer Yudkowsky on "Three Major Singularity Schools" our next speaker is eliezer yudkowsky he's from my colleagues at the singularity Institute he's one of the co-founders and research fellow he's going to be talking a little bit about the different schools of thoughts around this notion of the singularity and Eliez...
51122ea8-70a9-4a51-a7ec-88f7540005ed
trentmkelly/LessWrong-43k
LessWrong
Userscript to always show LW comments in context vs at the top The out of context top-level display of comments when I navigate to them always bothered me, but up until recently I haven't realized there is a way to go to the actual comment via a simple URL change.  From https://www.lesswrong.com/posts/<postid>/?comme...
23dc0b0a-1057-43c5-b640-43617308b138
trentmkelly/LessWrong-43k
LessWrong
What if Alignment is Not Enough? The following is a summary of Substrate Needs Convergence, as described in The Control Problem: Unsolved or Unsolvable?, No People as Pets (summarized here by Roman Yen), my podcast interview with Remmelt, and this conversation with Anders Sandberg.  Remmelt assisted in the editing of ...
da4d667c-2e09-49e0-9be1-02e47ccff55b
trentmkelly/LessWrong-43k
LessWrong
Zen and Rationality: Trust in Mind This is post 2/? about the intersection of my decades of LW-style rationality practice and my several years of Zen practice. In today's installment, I look at "trust in mind" from a rationalist perspective. I really like Trust in Mind, a Zen teaching poem attributed to the Third Pa...
d144a11b-04af-482c-a261-cce5fecad733
trentmkelly/LessWrong-43k
LessWrong
Amazon closing AmazonSmile to focus its philanthropic giving to programs with greater impact I just received the following email. There's also a notice up on their website. Since many folks here use AmazonSmile to make additional donations to charities, thought folks should know. > Dear customer, > > In 2013, we lau...
8a8b8253-6dc7-414c-ad6b-f049425724b9
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Caelum est Conterrens: I frankly don't see how this is a horror story So Eliezer said in his March 1st [HPMOR progress report](http://hpmor.com/notes/progress-13-03-01/): > > I recommend the recursive fanfic “[Friendship is Optimal: Caelum est Conterrens](http://www.fimfiction.net/story/69770/friendship-is-optimal...
2e96ee3a-ecd5-4659-b63d-e10ab9688a71
StampyAI/alignment-research-dataset/arxiv
Arxiv
Transfer of Adversarial Robustness Between Perturbation Types 1 Introduction --------------- Deep networks have shown remarkable accuracy on benchmark tasks (He et al., [2016](#bib.bib12)), but can also be fooled by imperceptible changes to inputs, known as adversarial examples (Goodfellow et al., [2014](#bib.bib10...
c94e2f11-dc76-4b93-b52c-d8bac140e4f8
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
[Linkpost] A General Language Assistant as a Laboratory for Alignment This is a linkpost for a recent paper from Anthropic: [A General Language Assistant as a Laboratory for Alignment](https://arxiv.org/abs/2112.00861). Abstract: > Given the broad capabilities of large language models, it should be possible to work...
d71d0fc9-0e37-4388-a325-91bd40b528a9
trentmkelly/LessWrong-43k
LessWrong
Convincing All Capability Researchers This post is heavily based off this excellent comment (the author's name is not relevant) > Give the world's thousand most respected AI researchers $1M each to spend 3 months working on AI alignment, with an extra $100M if by the end they can propose a solution alignment research...
3b1f2146-8c0e-4ccc-b717-3d342ed6f0d7
trentmkelly/LessWrong-43k
LessWrong
Killing Moloch with "Nudge tech"? Context I come from the SaaS startup world and I intend to help kill some inadequate equilibria in the next few years, specifically around the attention economy. I'll outline what I'm thinking, below, and would appreciate any help from the LW community: Specific questions I hope to ...
1f210dc2-4e71-414d-b8ef-19644f290ac1
trentmkelly/LessWrong-43k
LessWrong
Huntington's Disease; power and duty of parents over offsprings Huntington's Disease Basic facts about Huntington's Disease: * Huntington's Disease is an inherited disorder that results in death of brain cells. * Symptoms usually begin between 30 and 50 years of age, death typically occurs fifteen to twenty years l...
0d598391-80a8-4920-9dda-96f9ab5a05bd
trentmkelly/LessWrong-43k
LessWrong
3-bit filters I've come to appreciate my gut feeling for how things in my everyday life feel over time. To distill the process of checking in with your actual felt sense of things into a neat information-theoretic package, I've recently been using a 3-bit filter: Three yes-or-no questions you can ask yourself that wi...
f5bb1080-fd44-409c-a497-a1ac3e85e945
trentmkelly/LessWrong-43k
LessWrong
Estimating the benefits of a new flu drug (BXM) Introduction H5N1 is a looming threat, making regular headlines. The CDC has identified 66 U.S. human cases as of January 4, 2024. Scott Alexander has a recent post outlining the situation. There are several FDA-approved prescription antiviral flu drugs. The newest, usi...
d382f163-e359-4fef-887b-8003947adc15
trentmkelly/LessWrong-43k
LessWrong
Lawrence Watt-Evans's Fiction One of my pet topics, on which I will post more one of these days, is the Rationalist in Fiction.  Most of the time - it goes almost without saying - the Rationalist is done completely wrong.  In Hollywood, the Rationalist is a villain, or a cold emotionless foil, or a child who has to gr...
ce0ce27c-30fd-4803-b605-e53ec2ff636d
trentmkelly/LessWrong-43k
LessWrong
Meetup : Melbourne Social Meetup: June Discussion article for the meetup : Melbourne Social Meetup: June WHEN: 02 June 2017 06:30:00PM (+1000) WHERE: The Bull and Bear Tavern, Flinders Lane, Melbourne The MelbLW social meetup has moved to an exciting new schedule! We are now on the first Friday of each month. Come ...
e50d99d5-5225-4cf7-b866-aaa39f06be12
trentmkelly/LessWrong-43k
LessWrong
Loebian cooperation in the tiling agents problem The tiling agents problem is about formalizing how AIs can create successor AIs that are at least as smart. Here's a toy model I came up with, which is similar to Benya's old model but simpler. A computer program X is asked one of two questions: * Would you like some ...
97d1031f-711a-47c9-9c11-9ae41b923004
trentmkelly/LessWrong-43k
LessWrong
Meetup : Washington, D.C.: Fun & Games Discussion article for the meetup : Washington, D.C.: Fun & Games WHEN: 13 March 2016 02:00:00PM (-0500) WHERE: Reynolds Center We will be meeting to hang out, play games, and engage in fun conversation. If you would like to play a game that takes a long time, please feel wel...
8b297211-0910-4498-ba80-2fc34c5c4fb8
trentmkelly/LessWrong-43k
LessWrong
Yet more "stupid" questions This is a thread where people can ask questions that they would ordinarily feel embarrassed for not knowing the answer to. The previous thread is at close to 500 comments.
401f9535-eb94-45f6-a17b-a05ea4ad4245
trentmkelly/LessWrong-43k
LessWrong
[$10k bounty] Read and compile Robin Hanson’s best posts I think Robin Hanson's ideas are not read nearly as widely as they should be, in part because it's difficult to navigate his many, many blog posts (I estimate he's written 2000 of them exactly 3302 of them). So I'd like to pay someone to read through all his wri...
e5df8bb8-d1ab-48f4-b090-55b1b6de74ed
trentmkelly/LessWrong-43k
LessWrong
Randomness in Science I.  Humans have a randomness problem. We are bad at generating randomness as individuals (try to out-random a computer in Man vs. Machine) and in the aggregate - ask a group of people to choose a “random” number between 1-20 and the most common number will be 17. We are also bad at detecting ran...
1accf325-a1db-4435-9a85-b6c2583d5e92
trentmkelly/LessWrong-43k
LessWrong
Future directions for narrow value learning Narrow value learning is a huge field that people are already working on (though not by that name) and I can’t possibly do it justice. This post is primarily a list of things that I think are important and interesting, rather than an exhaustive list of directions to pursue. ...
a2091c00-bfa3-4b7a-b39b-8054b9cc9c7a
trentmkelly/LessWrong-43k
LessWrong
[Link Post] Bytes Are All You Need: Transformers Operating Directly On File Bytes > Modern deep learning approaches usually transform inputs into a modality-specific form. For example, the most common deep learning approach to image classification involves decoding image file bytes into an RGB tensor which is passed...
f4b207d3-352e-4207-9f76-5652fb9f5123
trentmkelly/LessWrong-43k
LessWrong
Intermittent Distillations #3 Mundane solutions to exotic problems (Paul Christiano) Mundane solutions to exotic problems Summary Thinking about AI safety often leads to considering exotic problems: models purposefully altering their gradients, agents hiding their capabilities to defect when an opportunity arises, o...
041195a4-781a-4c72-8247-313072724f92
trentmkelly/LessWrong-43k
LessWrong
AI Alignment Problem: Requirement not optional (A Critical Analysis through Mass Effect Trilogy) Author: TAWSIF AHMED Alignment, (Dictionary definition: a position of agreement or alliance). In Artificial Intelligence, Alignment means computers and human beings going along with each other “fine”.   Now coming back t...
5b5e7521-b62c-4294-b67e-a7cbcd395a50
trentmkelly/LessWrong-43k
LessWrong
Paper: Identifying the Risks of LM Agents with an LM-Emulated Sandbox - University of Toronto 2023 - Benchmark consisting of 36 high-stakes tools and 144 test cases! Paper: https://arxiv.org/abs/2309.15817 Github: https://github.com/ryoungj/toolemu Website: https://toolemu.com/ Abstract: > Recent advances in Langu...
44eb34e2-60ed-435c-984c-950aee289177
trentmkelly/LessWrong-43k
LessWrong
I Would Have Solved Alignment, But I Was Worried That Would Advance Timelines The alignment community is ostensibly a group of people concerned about AI risk. Lately, it would be more accurate to describe it as a group of people concerned about AI timelines. AI timelines have some relation to AI risk. Slower timeline...
7f959ce7-576b-4ff5-a061-aa70fc6b1264
StampyAI/alignment-research-dataset/blogs
Blogs
🐶Safetensors audited as really safe and becoming the default Audit shows that safetensors is safe and ready to become the default ==================================================================== [Hugging Face](https://huggingface.co/), in close collaboration with [EleutherAI](https://www.eleuther.ai/) and [Stab...
40bca758-9d1f-4824-b359-98a7348803a6
trentmkelly/LessWrong-43k
LessWrong
Meetup : Fort Collins, Colorado Meetup Wednesday Sept. 21, 7pm Discussion article for the meetup : Fort Collins, Colorado Meetup Wednesday Sept. 21, 7pm WHEN: 21 September 2011 07:00:00AM (-0600) WHERE: Bean Cycle, 144 North College Avenue, Fort Collins, CO 80524 An opportunity to practice improving your thinking s...
22626131-2187-4829-8c03-8e56f7c922a6
trentmkelly/LessWrong-43k
LessWrong
[Resolved] Who else prefers "AI alignment" to "AI safety?" Edit:  1. User davidad wrote a comprehensive overview of other actors in the field who have begun using AI alignment instead of AI safety as the standard terminology to refer to the control problem. It seems this is a general or growing preference in the fie...
6798b7d8-c12e-42eb-8ddc-ac5428a6b889
trentmkelly/LessWrong-43k
LessWrong
Effective Evil Many years ago, a blogger made a post advocating for an evil Y-Combinator which subsidized the opposite of Effective Altruism. Everyone (including the blogger) thought the post was a joke except the supervillains. The organization they founded celebrated its 10th anniversary this year. An attendee leake...
33f5a35a-bb6e-4551-b124-0dfc301b7a76
trentmkelly/LessWrong-43k
LessWrong
Pause AI Development? I like the basic idea of a pause in training increasingly powerful AIs. Yet I'm quite dissatisfied with any specific plan that I can think of. AI research is proceeding at a reckless pace. There's massive disagreement among intelligent people as to how dangerous this is. I participated in a for...
1c68a7fd-bad4-482b-9eac-8e6d59fa38a8
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Proofs Section 2.2 (Isomorphism to Expectations) .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: now...
672d9623-9de6-431f-8f78-bf43d07ae0b2
trentmkelly/LessWrong-43k
LessWrong
[LINK] SMBC on Confirmation Bias The latest SMBC made me laugh a bit, so I thought I'd bring extra LessWrong attention to it. I don't know if pointing out the subject of the comic in advance will make it more or less funny. Knowing that might be more data regarding that recent study claiming that spoilers don't actua...
a67b17ae-4174-415e-a3cd-149201a17583
trentmkelly/LessWrong-43k
LessWrong
Changes in Reality [Some short thoughts I just wanted to get out of my brain; bullet-points instead of well-structured prose. This is entirely random speculation, and not well-explained. Cross-posted from Grand, Unified, Empty.] * Social systems (laws, customs, memes) are subject to evolutionary pressure from the dy...
d9fd3a99-6e8b-455f-9963-cff1a8b9f2ba
trentmkelly/LessWrong-43k
LessWrong
Manifold Predicted the AI Extinction Statement and CAIS Wanted it Deleted  Crosspost from https://news.manifold.markets/p/manifold-predicted-the-ai-extinction History may look back on CAIS’s AI risk statement as a pivotal moment IF we survive AI. But, what if I told you that AI researchers “leaked” info about the st...
244547b7-8319-435d-a620-136124205980
trentmkelly/LessWrong-43k
LessWrong
Moral intuitions are surprisingly variable When building moral frameworks, moral intuitions are “our first and only source of data”. Further quoting the Consequentalism FAQ: > Searching for moral rules means searching for principles that correctly describe and justify enough of our existing moral intuition that we fe...
e0de5047-8e3a-45fd-af80-6f71a75759f4
trentmkelly/LessWrong-43k
LessWrong
Meetup : West LA—R:AZ Part D, Mysterious Answers Discussion article for the meetup : West LA—R:AZ Part D, Mysterious Answers WHEN: 22 April 2015 07:00:00PM (-0700) WHERE: 11066 Santa Monica Blvd, Los Angeles, CA How to Find Us: Go into this Del Taco. We will be in the back room if possible. Parking is free in the ...
22037fd8-1e7c-414d-8d06-25a30e660729
StampyAI/alignment-research-dataset/lesswrong
LessWrong
RLHF I’ve been thinking about Reinforcement Learning from Human Feedback (RLHF) a lot lately, mostly as a result of [my AGISF capstone project](https://github.com/anshradh/trl_custom) attempting to use it to teach a language model to write better responses to Reddit writing prompts, a la [Learning to summarize from hu...
c8225e56-7ebe-4155-9e14-c0abe22899fe
trentmkelly/LessWrong-43k
LessWrong
Environments for Measuring Deception, Resource Acquisition, and Ethical Violations Models are rapidly being deployed in the real world. How do we evaluate models, especially ones as complex as GPT-4, to ensure that they behave safely in pursuit of their objectives? Can we design models that robustly avoid any harms wh...
1e93272c-0984-4505-b331-6f2c6aac8d0d
trentmkelly/LessWrong-43k
LessWrong
Improving productivity and wellbeing Epistemic status : A decently thought out synthesis of a few books and other sources mixed together with my own thinking. This is not professional work, does not reflect personal knowledge of an expert consensus, and was not yet fully tested. For more on the limitations and flaws o...
f20e4502-57ac-49d8-a4a0-1e888a9b5e41
trentmkelly/LessWrong-43k
LessWrong
Assuming LK99 or similar: how to accelerate commercialization? Tabling the question of whether LK99 itself is a room temperature superconductor, I wonder about how we would commercialize it assuming it (or something similar) is, as rapidly as possible. I was reading Ben Reinhardt's Metalessons from the LK-99 Saga post...
062f0d98-c7af-4d90-b9c4-17b3c9c69e09
trentmkelly/LessWrong-43k
LessWrong
2021 Darwin Game - Contestants 123 players submitted a total of 556 species. Most players submitted just 1 species. One player submitted 11. (I disqualified his[1] 11ᵗʰ submission.) 555 species qualified for entry. I expected more players would submit 10 species compared to 8 or 9 in order to max out their submissions...
f2f32aff-cd44-4feb-b1c9-5eaa601fb438
trentmkelly/LessWrong-43k
LessWrong
On Claude 3.0 CLAUDE 3.0 Claude 3.0 is here. It is too early to know for certain how capable it is, but Claude 3.0’s largest version is in a similar class to GPT-4 and Gemini Advanced. It could plausibly now be the best model for many practical uses, with praise especially coming in on coding and creative writing. A...
e9ea1dbb-cf0e-474a-8069-15cee3d3c1e1
trentmkelly/LessWrong-43k
LessWrong
Prices or Bindings? Followup to:  Ethical Injunctions During World War II, Knut Haukelid and three other saboteurs sank a civilian Norwegian ferry ship, the SF Hydro, carrying a shipment of deuterium for use as a neutron moderator in Germany's atomic weapons program.  Eighteen dead, twenty-nine survivors.  And that w...
01b80cd5-6e3d-45d6-86c7-38ae20984213
trentmkelly/LessWrong-43k
LessWrong
On pointless waiting I’ve often noticed in myself a tendency, if I am not doing something immediately engrossing, to find myself waiting. Waiting, waiting, waiting, not really being present, just willing time to pass. But the weird thing is, frequently there isn’t anything in particular that I’m waiting for. Getting...
688a6b4d-fb27-432e-94a6-e2589abb93bf
StampyAI/alignment-research-dataset/arxiv
Arxiv
Informed Machine Learning -- A Taxonomy and Survey of Integrating Knowledge into Learning Systems 1 Introduction --------------- Machine learning has shown great success in building models for pattern recognition in domains ranging from computer vision [[1](#bib.bib1)] over speech recognition [[2](#bib.bib2)] and t...
c9432e54-a1ee-45a2-beaf-da13311e8580
trentmkelly/LessWrong-43k
LessWrong
What is the Main/Discussion distinction, and what should it be? Near the beginning of this year Wei Dai asked why certain people don't post to LessWrong more often, and Yvain replied that: > Less Wrong requires no politics / minimal humor / definitely unambiguously rationality-relevant / careful referencing / airtigh...
5d1d7754-d854-44ec-b4ec-cb2af7f5a867
trentmkelly/LessWrong-43k
LessWrong
Identifiability Problem for Superrational Decision Theories Superrationality, and generalizations of it, must treat options differently depending on how they're named. Consider the penny correlation game: Both players decide independently on either head or tails. Then if they decided on the same thing, they each get ...
94150b5d-de4e-44ae-8289-78f961d2d684
trentmkelly/LessWrong-43k
LessWrong
Every "best paper" from Computer Science conferences since 1996 [link] http://jeffhuang.com/best_paper_awards.html http://news.ycombinator.com/item?id=2051437
4cc02e5d-8c6b-4935-b83d-3d0b1e4b8d62
trentmkelly/LessWrong-43k
LessWrong
Metauncertainty Response to: When (Not) To Use Probabilities “It appears to be a quite general principle that, whenever there is a randomized way of doing something, then there is a nonrandomized way that delivers better performance but requires more thought.” —E. T. Jaynes The uncertainty due to vague (non math) la...
1c42b8f6-2dd8-423f-8082-7d9d72286055
trentmkelly/LessWrong-43k
LessWrong
What Comes After Epistemic Spot Checks? When I read a non-fiction book, I want to know if it’s correct before I commit anything it says to memory. But if I already knew the truth status of all of its claims, I wouldn’t need to read it. Epistemic Spot Checks are an attempt to square that circle by sampling a book’s cla...
191cf186-3635-4311-840d-3961d385a5b0
StampyAI/alignment-research-dataset/arxiv
Arxiv
Forecasting using incomplete models 1 Introduction --------------- Forecasting future observations based on past observations is one of the fundamental problems in machine learning, and more broadly is one of the fundamental components of rational reasoning in general. This problem received much attention, using di...
916fbad2-0f78-4cfe-993e-2f361071ba96
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Modeling the impact of AI safety field-building programs Summary ======= AI safety field-building is a large space, with many possible programs. The [Center for AI Safety](https://www.safe.ai/) (CAIS) has commissioned cost-effectiveness models to support decisions regarding how to prioritize between these programs. D...
5cbcc48c-ab47-41cd-8dd0-24842c97252c
trentmkelly/LessWrong-43k
LessWrong
Unnatural Categories Followup to:  Disguised Queries, Superexponential Conceptspace If a tree falls in the forest, and no one hears it, does it make a sound? "Tell me why you want to know," says the rationalist, "and I'll tell you the answer."  If you want to know whether your seismograph, located nearby, will regis...
74b47a22-099a-47ad-aaae-d31f82b80edb
trentmkelly/LessWrong-43k
LessWrong
[Paper] Simulation of a complete cell "A Whole-Cell Computational Model Predicts Phenotype from Genotype" by Jonathan Karr et al. This paper appeared a few days ago in Cell, and describes a computational simulation of the bacterium Mycoplasma genitalium, conducted at this lab. The paper is behind a paywall, but is bl...
ddaff4a6-3dd0-405b-91b6-1fe24d036b13
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post276 (thanks to Alex Mallen, Cody Rushing, Zach Stein-Perlman, Hoagy Cunningham, Vlad Mikulik, and Fabien Roger for comments) Sometimes I hear people argue against AI control as follows: if your control measures rely on getting good judgments from "trusted" AI models, you're doomed because AI models aren't adve...
5d31da87-32cf-4b59-8ed1-1ff92fd1dd47
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
The future of digitalization (Gerhard Fischer) - 1st AiTech Symposium so I think they're organizers for inviting me and so is this is the title of my talk and as Lisa mentioned already one of my of the interest for me has been design trade-offs and understanding what that means and so I want to start with sort of this...
608e55f7-5d97-42b6-b759-872f99e0bb0e
trentmkelly/LessWrong-43k
LessWrong
Meetup : Sydney Rationality Dojo - September 2016 Discussion article for the meetup : Sydney Rationality Dojo - September 2016 WHEN: 04 September 2016 04:00:00PM (+1000) WHERE: 10 Shepherd Street, Chippendale From 4 - 6 we'll be holding our normal rationality dojo at Humanist house. Afterwards there'll be an inform...
b7f637d8-651a-411c-aa37-452557ec7f69
trentmkelly/LessWrong-43k
LessWrong
Meetup : Munich April Meetup Discussion article for the meetup : Munich April Meetup WHEN: 18 April 2015 03:00:03PM (+0200) WHERE: Café Puck, Türkenstraße 33, Munich It's time for another meetup! If you are new, please do come and say hi. Topics: Let's develop and play a prediction game, taking some inspirations f...
6182a535-354b-44e2-894c-bb1a95b36a1b
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Brain-Computer Interfaces and AI Alignment *cross-posted from [niplav.github.io](https://niplav.github.io)* > > I discuss arguments for and against the usefulness of brain-computer > interfaces in relation to AI alignment, and conclude that the path to > AI going well using brain-computer interfaces hasn't been ex...
6f65b9f7-8e77-4125-8c94-58c9904ae850
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Thoughts on AI Safety Camp **I** Early this year I interviewed a sample of AISC participants and mentors, and spent some time thinking about the problems the AI safety research community is facing, and have changed my mind about some things. [AI Safety Camp](https://aisafety.camp/) is a program that brings together ...
4afced18-912d-4d97-8fba-723f9dcc7ab1
trentmkelly/LessWrong-43k
LessWrong
Has anyone thought about how to proceed now that AI notkilleveryoneism is becoming more relevant/is approaching the Overton window? I figure that this would be a pretty important question since getting governments on board, and getting people to coordinate in general, could buy us a lot of time, but I haven't seen any...
ba635d9b-f02f-4195-b8a1-cf81150aae72
trentmkelly/LessWrong-43k
LessWrong
Change A View: An interesting online community I've recently stumbled across the platform Change A View, which is a subreddit that forked/graduated to its own website that is centered around the idea of people creating posts in which they ask to have their mind changed. This tends to create a frame that enables much m...
bd0b3832-6a2a-47e9-90c5-e933f6b1ad5d
trentmkelly/LessWrong-43k
LessWrong
Sleeping Beauty: an Accuracy-based Approach This post does not propose a solution to the Sleeping Beauty problem, but presents arguments based on accuracy for thirders, halfers and double-halfers. A more detailed draft paper can be found here. Summary Accuracy-based arguments claim that one should plan to adopt post...
20819746-b964-49ad-8c1b-d917fb448cf9
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Fake Norms, or "Truth" vs. Truth Today's post, Fake Norms, or "Truth" vs. Truth was originally published on 22 July 2008. A summary (taken from the LW wiki):   > Our society has a moral norm for applauding "truth", but actual truths get much less applause (this is a bad thing). Discuss the post here (r...
c6860eb8-0530-4b78-9401-0f8a8e00a9e0
trentmkelly/LessWrong-43k
LessWrong
Ablations for “Frontier Models are Capable of In-context Scheming” We recently published our paper “Frontier Models are Capable of In-context Scheming”. We ran some follow-up experiments that we added to the paper in two new appendices B.5 and B.6. We summarize these follow-up experiments in this post. This post assum...
b4a555df-5d78-4570-b5f2-1f57e5cdc661
trentmkelly/LessWrong-43k
LessWrong
Making a New Table Leaf We got our dining room table used, and it didn't come with any leaves. I made one out of plywood when we first got it, and it's been pretty helpful over the past ~7y: With the leaf we can comfortably fit eight: three on the long sides and one on the short. If we have more people two on the s...
9e9e22a5-44b3-483a-8dad-987630f3e774
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Charbel-Raphaël and Lucius discuss Interpretability The following is a transcript of a public discussion between [Charbel-Raphaël Segerie](https://www.lesswrong.com/users/charbel-raphael) and [Lucius Bushnaq](https://www.lesswrong.com/users/lblack) that took place on 23 September during [LessWrong Community Weekend 20...
09c33d89-51cf-483f-9630-623be740d71a
trentmkelly/LessWrong-43k
LessWrong
Progress links and tweets, 2023-05-16 Opportunities * Animator needed for a molecular biology illustration project (via @kanzure) * Heidi Williams seeking a biotech VC co-author on improving university tech transfer Announcements * Foresight Vision Weekend 2023 (via @foresightinst). I’ll be speaking! * Helion ha...