id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
64156c47-8c42-468f-bfd7-f1d50f23000f | trentmkelly/LessWrong-43k | LessWrong | Could LLMs Learn to Detect Bias Autonomously, Like Tesla’s Self-Driving Cars?
LLMs are shaped by human curator making decisions about what data is “ethical” or “safe.” This curation introduces biases, focusing on politically salient issues like fairness for marginalized groups while often overlooking subtler distortio... |
529062d2-8696-4ac9-a3e0-0b631f3d26ca | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Steering AI to care for animals, and soon
I'm worried that animal welfare advocates might neglect the importance of AI in determining what happens to animals. More specifically, I'm worried that the value
* {animal welfare matters}
and the true-according-to-me belief
* {AI is going to transform and determine mos... |
f02730ef-643d-4f7d-8de1-17c9acf98633 | trentmkelly/LessWrong-43k | LessWrong | Generalising CNNs
TL/DR; The invention of Convolution Neural Networks for image and audio processing was a key landmark in machine learning.
This topic is for people who already know what CNNs are, and are interested in how to innovate to riff on and extend the core reason (perhaps?) that CNNs learn faster. Probing t... |
5dd025a7-179a-4f45-879b-cf882a98e722 | trentmkelly/LessWrong-43k | LessWrong | Latacora might be of interest to some AI Safety organizations
Am I affiliated with Latacora?
No, not in any way.
What is Latacora?
From their own webpage:
> Latacora does just one kind of engagement: we join your engineering team virtually, create a security practice, and run it. As your company matures, we'll make... |
ff939eff-f11c-472d-babb-45ce1c3d8505 | trentmkelly/LessWrong-43k | LessWrong | Are SAE features from the Base Model still meaningful to LLaVA?
Shan Chen, Jack Gallifant, Kuleen Sasse, Danielle Bitterman[1]
Please read this as a work in progress where we are colleagues sharing this in a lab (https://www.bittermanlab.org) meeting to help/motivate potential parallel research.
TL;DR:
* Recent wor... |
84672661-5d62-4728-9743-f757d9fbfd93 | trentmkelly/LessWrong-43k | LessWrong | Learning human preferences: black-box, white-box, and structured white-box access
This post is inspired by system identification; however, I'm not an expert in that domain, so any corrections or inspirations on that front are welcome.
I want to thank Rebecca Gorman for her idea on using system identification, and her... |
60318780-3456-4cee-9bd2-f994f8cde5c1 | trentmkelly/LessWrong-43k | LessWrong | [LINK] Ethical Pick-Up Artistry (Clarisse Thorn)
Clarisse Thorn recently posted a useful article about Ethical Pick-Up Artistry, bringing up a few basic critiques of traditional PUA and suggesting a few alternatives.
> Here’s the thing: the current pickup artist subculture has a monopoly on effective advice for how t... |
778aa42b-0fae-40f7-80dd-77c070384865 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | [ASoT] Natural abstractions and AlphaZero
I just read [Acquisition of Chess Knowledge in AlphaZero](https://arxiv.org/abs/2111.09259) and it's both really cool and has interesting implications for the Natural Abstractions Hypothesis. AZ was trained with *no human data* and yet it settles on relatively interpretable ab... |
ba32ca0b-64d8-4d26-ad5b-b44e8468b6ee | StampyAI/alignment-research-dataset/arbital | Arbital | Subgroup is normal if and only if it is the kernel of a homomorphism
Let $N$ be a [https://arbital.com/p/-subgroup](https://arbital.com/p/-subgroup) of [https://arbital.com/p/-3gd](https://arbital.com/p/-3gd) $G$.
Then $N$ is [normal](https://arbital.com/p/4h6) in $G$ if and only if there is a group $H$ and a [https:/... |
cf47adaf-7de1-4de8-81cc-5ef1ef9e4e82 | StampyAI/alignment-research-dataset/blogs | Blogs | Evolution through large models
Evolution through Large Models
Joel Lehman
OpenAI
joel@openai.comJonathan Gordon
OpenAI
gordonjo@openai.com
Shawn Jain
OpenAI
jains@openai.comKamal Ndousse
Anthropic∗
kamal.ndousse@gmail.com
Cathy Yeh
OpenAI
cathy@openai.comKenneth O. Stanley
OpenAI
kennethostanley@gmail.com
June 20, 202... |
dfb3532a-6e9b-4c17-b219-a1295a3508b4 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Less Wrong NH Meet-up
Discussion article for the meetup : Less Wrong NH Meet-up
WHEN: 10 November 2015 07:00:00PM (-0500)
WHERE: 269 Pearl St Manchester NH 03104
The seventh NH meet-up is Tuesday, 11/10, in Manchester, NH at 7 pm at a private residence. Light refreshments will be provided.
Have you read R... |
7847de9b-26e0-44bf-93b4-a8af9707d33e | trentmkelly/LessWrong-43k | LessWrong | Comp Sci in 2027 (Short story by Eliezer Yudkowsky)
Comp sci in 2017:
Student: I get the feeling the compiler is just ignoring all my comments.
Teaching assistant: You have failed to understand not just compilers but the concept of computation itself.
Comp sci in 2027:
Student: I get the feeling the compiler is... |
1ad9683b-0a9c-4e23-b094-0e00027b8343 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "If you want to understand the modern economy, as opposed to the economies of yore, one source I strongly recommend is a short story from the July 1958 issue of Astounding Science Fiction, titled “Business As Usual During Alterations”. It’s roughly a 15 minute read. I’m about to throw out major spoilers, so stop readin... |
4b629264-4a86-4dab-98b2-ece1ad839a6a | trentmkelly/LessWrong-43k | LessWrong | Meetup : San Francisco Meetup: Group Debugging
Discussion article for the meetup : San Francisco Meetup: Group Debugging
WHEN: 14 November 2016 06:15:00PM (-0800)
WHERE: 1769 15th St, San Francisco, CA
We’ll be meeting to solve each other’s problems!
Here’s how it works: we’ll have everyone brainstorm silently for... |
7a2ca354-0921-42fb-9928-590f84774ce8 | StampyAI/alignment-research-dataset/arxiv | Arxiv | How can AI Automate End-to-End Data Science?
1 Introduction and Motivation
------------------------------
Data science covers the whole spectrum of data processing, beginning from data integration, distributed architecture, automating machine learning, data visualization, dashboards and BI, data engineering, deploy... |
fcf32ee5-38ed-4afb-939a-c19531ceb1b8 | trentmkelly/LessWrong-43k | LessWrong | Inflection AI: New startup related to language models
Inflection AI (website, announcement) is a new startup founded by Mustafa Suleyman, Reid Hoffman, and Karén Simonyan. It appears they'll be doing something with language models. They're also looking to hire a safety team; I bet they'll need to do RLHF as well as to... |
c24e96fa-21e8-40ee-a2e2-e2eab44c89be | StampyAI/alignment-research-dataset/arxiv | Arxiv | Image Augmentation Is All You Need: Regularizing Deep Reinforcement Learning from Pixels
1 Introduction
---------------
Sample-efficient deep reinforcement learning (RL) algorithms capable of directly training from image pixels would open up many real-world applications in control and robotics. However, simultaneou... |
df5ebf34-e488-4831-8e71-9f27f9bdb592 | trentmkelly/LessWrong-43k | LessWrong | Approach to Screen Time
Screens can be really engrossing, and learning moderation is pretty hard. This is one of a small number of places (along with staying up as late as they want and eating as many sweets as they'd like) where an approach of letting our kids play with them as much as they want doesn't seem like it ... |
79fd2b7d-48e6-4d0d-91d4-626b0349d8b6 | trentmkelly/LessWrong-43k | LessWrong | Fighting in various places for a really long time
The first time someone raved to me about seeing Everything Everywhere All at Once, I thought they were actually suggesting I see everything everywhere all at once, and I was briefly excited by the implication that this exhilarating possibility was somehow on the table.... |
dc4ef93f-47a4-462e-ac7d-8c25542d75fd | trentmkelly/LessWrong-43k | LessWrong | You're Not One "You" - How Decision Theories Are Talking Past Each Other
(Crossposted from MidWittgenstein (as part of my lame resolution to write more this year))
This is a post about what I see as a bit of a confusion in some Decision Theory arguments. It touches on FDT criticisms of CDT/EDT so I wanted to post her... |
df676576-3dcb-40b4-9687-280714031b78 | StampyAI/alignment-research-dataset/blogs | Blogs | AI alignment timeline codes
AI alignment timeline codes
---------------------------
this is a small post proposing simple one-letter codes for identifying timelines depending on their status relative to [AI alignment](https://en.wikipedia.org/wiki/AI_control_problem) and the appearance of [superintelligence](https:/... |
c1dd53c4-c8ed-43da-9e02-0e2b3afe679b | trentmkelly/LessWrong-43k | LessWrong | Book Review: Radical Markets
This is a book review of Radical Markets, taken from my blog Random Walks. I summarize and critique each chapter to varying extents. I give the book a rating of 2/5, and recommend reading the first two chapters.
----------------------------------------
Radical Markets? More like Activist... |
d25b1564-16e9-4103-93ae-b179bc18668a | trentmkelly/LessWrong-43k | LessWrong | The Black Team - A Parable of Group Effectiveness
> Management noticed that certain software testers were 10 to 20 percent better at finding defects than their peers. By putting these people on the same team, they reasoned, they could form a group that would be 10 or 20 percent more effective and then put the team to ... |
f795adaa-209e-4b84-bea7-f1b212dd1b44 | trentmkelly/LessWrong-43k | LessWrong | The manifest manifesto
Followup to: If you weren't such an idiot...
We believe in staying alive — at least when life is or will be worth it — to do things that are useful and/or fun.
Protect yourself from extreme cold by techniques such as wearing clothes. Avoid looking directly at the sun, except when using eye-pro... |
f5913882-82e0-47cd-8672-fc8197a15ee6 | trentmkelly/LessWrong-43k | LessWrong | Another Concise Open Problem
Given the success of the last open problem I posted, (Janos Kramer disproved my conjecture) I decided to post another one. Again, I will cut off the philosophy, and just give the math. The reason I want this result is to solve the problem described here.
Let S be the set of all paris (M,k... |
449ba5c0-f980-4d55-b14e-06313b97249d | trentmkelly/LessWrong-43k | LessWrong | Three Levels of Motivation
Epistemic Status: This is basic information, and while it is original to me, I make no claim that the the content thereof is new. Nevertheless, I think it is important enough to be worth posting. |
e86ad5ac-060b-4e48-b923-82aa13e8f8f3 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Ngo and Yudkowsky on scientific reasoning and pivotal acts
This is a transcript of a conversation between Richard Ngo and Eliezer Yudkowsky, facilitated by Nate Soares (and with some comments from Carl Shulman). This transcript continues the [Late 2021 MIRI Conversations](https://intelligence.org/late-2021-miri-conver... |
0a07bfb1-9e39-41d9-b71f-0932c0f2b866 | trentmkelly/LessWrong-43k | LessWrong | Scientism vs. people
> The Benefactor, the ancestor of Orwell's Big Brother, is the absolute ruler of OneState. ... He rules over a human society that is deemed to have achieved, with only negligible exceptions, absolute perfection. Men have finally become, if not actually machines, as machine-like as possible, utterl... |
cdb28186-a35b-4c05-b87a-3f303d6dd7b7 | trentmkelly/LessWrong-43k | LessWrong | Posts with clickable sections of images?
I have an image of a diagram that I'd like to add hover-over or clickable hot spots that would pop up a window or take the reader to a footnote. Is this possible with the native functionality of the LW Site? |
e73b6be1-e73a-41e2-a436-64a32ad1431e | trentmkelly/LessWrong-43k | LessWrong | Just a reminder, for everyone that signed up for the intro to AI class, it's started.
Here you go: http://www.ai-class.com/
They're still working out kinks in the site, and no homework is up yet, but the initial set of lectures (really really basic "what is AI/welcome to the first day of class" type stuff is up), ... |
870d03fb-f3f3-4f5d-842b-34a6687b440e | trentmkelly/LessWrong-43k | LessWrong | Thermodynamic entropy = Kolmogorov complexity
Direct PDF link for non-subscribers
> Information theory must precede probability theory, and not be based on it. By the
> very essence of this discipline, the foundations of information theory have a finite combinatorial character.
>
> - Andrey Kolmogorov
Many alignme... |
b42902df-a202-441b-8b53-ed7497371967 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Pretraining Language Models with Human Preferences
1 Introduction
---------------
Conventional LM pretraining Pretraining with feedback
Finetuning with feedback for 1.6B tokens
Finetuning with feedback for 330M tokens
Figure 1... |
97e506dc-c84b-4d51-947f-9711fb0c4121 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Deterministic Policy Optimization by Combining Pathwise and Score Function Estimators for Discrete Action Spaces
Introduction
------------
Reinforcement and imitation learning using deep neural networks have achieved impressive results on a wide range of tasks spanning manipulation [levine2016end, levine2014learnin... |
7d2a18a5-367f-41f9-9241-562fd949af0c | trentmkelly/LessWrong-43k | LessWrong | Quantified Self recommendations
Is self knowledge through numbers worth the effort? I welcome your stories and recommendations. Thanks! |
95932d31-da6e-4ef4-a633-a9bde336064b | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "Followup to: The Useful Idea of Truth (minor post)
So far as I know, the first piece of rationalist fiction - one of only two explicitly rationalist fictions I know of that didn't descend from HPMOR, the other being "David's Sling" by Marc Stiegler - is the Null-A series by A. E. van Vogt. In Vogt's story, the protago... |
5cbef221-7b05-4b75-8a7f-58b8c795c4c3 | trentmkelly/LessWrong-43k | LessWrong | An analysis of the Less Wrong D&D.Sci 4th Edition game
This is an analysis of the game described in this post - you might want to read that first. Or if you don't want to, here's a quick summary: There's a team-based competitive video game, the Green team has a game coming up against the Blue team, and we're supposed ... |
9bf81cd8-8d7a-4e01-ba9b-c4840e2f6431 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | The Ubiquitous Converse Lawvere Problem
.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; floa... |
059ecb48-0755-448d-8fe6-e26b5b0acf74 | trentmkelly/LessWrong-43k | LessWrong | Love Island USA Season 7 Episode 20: What Could The Producers Be Thinking
(Note: This is NOT being posted on my Substack or Wordpress, but I do want a record of it that is timestamped and accessible for various reasons, so I'm putting this here, but do not in any way feel like you need to read it unless it sounds like... |
80354aeb-fa3c-4ee5-9d1f-b464f8de6ff7 | trentmkelly/LessWrong-43k | LessWrong | No - AI is just as energy-efficient as your brain.
This is simply because AIs run on electricity.
Our brains use sunlight via photosynthesis via dietary energy intake. From sunlight to dietary energy this is about 0.25-0.5% energy efficient. Let's say it's 0.35% efficient. (This is then our complete sunlight-to-capab... |
ed5874a3-610c-4249-a410-171e0810a14d | StampyAI/alignment-research-dataset/arbital | Arbital | Big-picture strategic awareness
Many [convergent instrumental strategies](https://arbital.com/p/10g) seem like they should arise naturally at the point where a [consequentialist](https://arbital.com/p/9h) agent gains a broad strategic understanding of its own situation, e.g:
- That it is an AI;
- Running on a compute... |
a57304d7-aaa3-46b9-b363-a0e8e115ba49 | trentmkelly/LessWrong-43k | LessWrong | [RESCHEDULED] NYC Rationality Megameetup and Unconference: 4/5 - 4/6
Please see the updated post here:
http://lesswrong.com/r/discussion/lw/jqh/rescheduled_nyc_rationality_megameetup_and/
On the weekend of April 5-6, the NYC community will be hosting a megameetup and rationality unconference. Everyone who can make t... |
770d2e5d-2c66-4bef-970e-d836826fbcf9 | trentmkelly/LessWrong-43k | LessWrong | Facets and Social Networks
It's rare that an author's interests will fully overlap with any given reader's. You can choose to post anywhere on a spectrum between:
* Narrow: pick a topic and make it your beat. People who care about that can follow you and pretty reliably see what they're looking for.
* Broad: wri... |
c1da8249-ba8e-47f6-af13-b76195cc927b | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | AI Benefits Post 5: Outstanding Questions on Governing Benefits
This is the final post in a series on "AI Benefits." It is cross-posted from my [personal blog](https://www.cullenokeefe.com/blog/ai-benefits-5). For other entries in this series, navigate to the [AI Benefits Blog Series Index page](https://www.cullenokee... |
29b7fce6-ca81-4f35-aeed-a7f3d556e341 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Sandy, UT - Debugging
Discussion article for the meetup : Sandy, UT - Debugging
WHEN: 19 December 2014 12:00:30PM (-0700)
WHERE: 9425 S Riverside Dr. Sandy, UT
Make friends, learn skills, and build solutions!
Meetup is in the main building, you can just walk in. Agenda: Quick Definitions - What is Life De... |
b214bf74-a87d-4228-80df-6b52e83375dc | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Twitter-length responses to 24 AI alignment arguments
Cross-posted from Twitter, on request:
(*Warning: arguments are Twitter-concise. Treat this all as a conversation-starter, rather than as anything authoritative or exhaustive.*)
---
"Bad take" bingo cards are terrible, because they never actually say what's wr... |
9d8da7d2-3280-44b4-96d6-f05153d81241 | trentmkelly/LessWrong-43k | LessWrong | Conditioning Predictive Models: Deployment strategy
This is the sixth of seven posts in the Conditioning Predictive Models Sequence based on the paper “Conditioning Predictive Models: Risks and Strategies” by Evan Hubinger, Adam Jermyn, Johannes Treutlein, Rubi Hudson, and Kate Woolverton. Each post in the sequence co... |
24687115-ea13-4fbb-9bd2-ccc56fdce742 | trentmkelly/LessWrong-43k | LessWrong | Any thoughts on how to locate job opportunities in Europe for US Citizen?
I am a graduate student in the U.S. nearing the completion of my master's degree in applied mathematics. I am looking for jobs at the master's degree level that involve working on large data sets, doing machine learning, scientific computing, et... |
aba6330b-e448-4c58-b0b7-90907b8594ef | trentmkelly/LessWrong-43k | LessWrong | Predict responses to the "existential risk from AI" survey
I sent a short survey to ~117 people working on long-term AI issues, asking about the level of existential risk from AI; 44 responded.
In ~6 days, I'm going to post the anonymized results. For now, I'm posting the methods section of my post so anyone interest... |
32b38d33-1bb4-4275-ac1c-3de5d7ef693e | trentmkelly/LessWrong-43k | LessWrong | Launching Adjacent News
Forward Looking News Driven by Prediction Markets and Forecasting Platforms
Prediction markets and forecasting platforms are an incredible news source, are interesting to trade, and provide tons of unique data:
1. News: Driving news by changes in market odds and providing context to news you... |
a95fa131-9647-46b4-92ef-ed963543bba7 | trentmkelly/LessWrong-43k | LessWrong | Summary: "How to Write Quickly..." by John Wentworth
This is part of my "research" series.
Wentworth, John S. How To Write Quickly While Maintaining Epistemic Rigor. LessWrong (blog), August 28, 2021. https://www.lesswrong.com/posts/Psr9tnQFuEXiuqGcR/how-to-write-quickly-while-maintaining-epistemic-rigor. Accessed 20... |
5abe80de-7b7d-4b76-a1f0-580c4b7c8311 | trentmkelly/LessWrong-43k | LessWrong | AGI safety from first principles: Superintelligence
In order to understand superintelligence, we should first characterise what we mean by intelligence. We can start with Legg’s well-known definition, which identifies intelligence as the ability to do well on a broad range of cognitive tasks.[1] The key distinction I’... |
f9bf993a-8e94-4925-87e9-6d2182aa81c4 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Polluting the agentic commons
The risks from developi... |
e8bb3294-be80-4dbb-9d16-0046e06cdf7c | trentmkelly/LessWrong-43k | LessWrong | D&D.Sci Hypersphere Analysis Part 3: Beat it with Linear Algebra
As before, this is composed entirely of spoilers for abstractapplic's D&D.Sci scenario, don't read if you want to solve on your own!
Last time, we left off with a solution that could reliably get us ~95% performance on all our generators. Sadly, in the... |
c0f0f940-5e52-4baf-9ed6-919f640b75b8 | trentmkelly/LessWrong-43k | LessWrong | College discussion thread
It's that time of year when high school seniors are thinking about colleges, and by extension, everyone who knows any high school seniors is thinking about colleges as well. So let's let Less Wrong join in!
Do you have:
* questions about choosing, preparing for, or attending colleges?
* s... |
de32dc1f-ff04-415f-a06e-42f34558e1b0 | trentmkelly/LessWrong-43k | LessWrong | AI #47: Meet the New Year
[NOTE: I forgot to post this to WP/LW/RSS on Thursday, so posting it now. Sorry about that.]
Will be very different from the old year by the time we are done. This year, it seems like various continuations of the old one. Sometimes I look back on the week, and I wonder how so much happened, ... |
cf03b991-613a-4834-9aee-2fe4e6f488df | StampyAI/alignment-research-dataset/blogs | Blogs | 2013 in Review: Strategic and Expository Research
This is the 3rd part of my personal and qualitative [self-review of MIRI in 2013](http://intelligence.org/2013/12/20/2013-in-review-operations/), in which I begin to review MIRI’s 2013 research activities. By “research activities” I mean to include outreach efforts pri... |
f93133f8-0a37-4b6c-8acd-4696a42327d4 | trentmkelly/LessWrong-43k | LessWrong | What are examples of Rationalist fable-like stories?
I will post some of the collection I've acquired. I have ideas for more. And would be interested in seeing more.
If there's enough interest, and with the permission of the authors, I would like to make a book of Rationalist fables / parables (but with a more generi... |
3800196e-97cd-469f-b3dc-d7d41a51cd39 | trentmkelly/LessWrong-43k | LessWrong | Existential Risk
This is a "basics" article, intended for introducing people to the concept of existential risk.
On September 26, 1983, Soviet officer Stanislav Petrov saved the world.
Three weeks earlier, Soviet interceptors had shot down a commercial jet, thinking it was on a spy mission. All 269 passengers were... |
e8279134-b848-4207-b89e-4e06023b5dda | StampyAI/alignment-research-dataset/lesswrong | LessWrong | GTP4 capable of limited recursive improving?
Apparently with reflection technique (answer-critique-improve) GTP4 capable of giving much better answers. But that implies it should be capable of doing essentially Alpha Go Zero type of learning! It can't do complete self play from zero as there is no ground truth for it ... |
2f629cdd-5a84-4c7c-9d2f-1a54233e5253 | trentmkelly/LessWrong-43k | LessWrong | Lessons from a year of university AI safety field building
This post is an organizational update from Georgia Tech’s AI Safety Initiative (AISI) and roughly represents our collective view. In this post, we share lessons & takes from the 2024-25 academic year, describe what we’ve done, and detail our plans for the next... |
67c7da4f-9c6a-4124-bd4c-94e7e64b3612 | trentmkelly/LessWrong-43k | LessWrong | A (paraconsistent) logic to deal with inconsistent preferences
One problem in AI/policy/ethics is that it seems like we sometimes have inconsistent preferences.[1] For example, I want to have a painting in my office, and simultaneously don’t want to have a painting in my office.[2] I’d both prefer and not prefer φ. Th... |
8f387706-9d7f-4d63-8f29-ee086db3cdb8 | trentmkelly/LessWrong-43k | LessWrong | #2: Neurocryopreservation vs whole-body preservation
This is post 2 of 10 in my sequence on how to sign up for cryonics.
Background
There are currently two options for human cryopreservation: preserving the whole body, or preserving only the brain (and the head that holds it). Note that while cryonics providers Alco... |
e5fa9452-1937-4e26-86e2-20e65945c8ec | trentmkelly/LessWrong-43k | LessWrong | The Rising Sea
And then we hit a wall.
Nobody expected it. Well... almost nobody. Yann LeCun posted his "I told you so's" all over X. Gary Marcus insisted he'd predicted this all along. Sam Altman pivoted, declaring o3 was actually already ASI.
The first rumors of scaling laws breaking down were already circulatin... |
84dd2c9b-f795-48ac-bb9c-69b33ff440d7 | trentmkelly/LessWrong-43k | LessWrong | Rare and Valuable skills worth mastering?
Hey everyone I just read this, which in turn lead me to this and I'm curious as to what are some skills and fields of study that people think fit the category of 'rare and valuable'.
Not just in white collar environments either, but also considering the entertainment industry... |
5950058a-a837-4e24-9dd0-aff15da6c8b5 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Berkeley meetup: Board games!
Discussion article for the meetup : Berkeley meetup: Board games!
WHEN: 09 January 2013 07:00:00PM (-0800)
WHERE: Berkeley, CA
Hello, all! Zendo will host its first Less Wrong meetup of the year on Wednesday at 7pm. Come and play a game from Zendo's library of board games, or ... |
aa9e08f2-425c-4dba-83b4-80fe34b16580 | trentmkelly/LessWrong-43k | LessWrong | Fake Journal Club proposal
|
7b02fae4-adf5-45d9-869c-4089c14a59a7 | trentmkelly/LessWrong-43k | LessWrong | Sleeping Beauty Problem Can Be Explained by Perspective Disagreement (I)
EDIT: It seems this is not getting much attention except for part II with the counter example to SIA. Let me try to intrigue you in to reading this. It provides a systematic argument for double halving in the sleeping beauty problem, if my reason... |
fa00e388-b277-4d7c-8c49-be36a3d112da | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "Prompted by Mitchell Porter, I asked on Physics StackExchange about the accuracy of the physics in the Quantum Physics sequence:
What errors would one learn from Eliezer Yudkowsky's introduction to quantum physics?
Eliezer Yudkowsky wrote an introduction to quantum physics from a strictly realist standpoint. However, ... |
698718c3-9996-4142-9a40-d0254abdc6fc | StampyAI/alignment-research-dataset/aisafety.info | AI Safety Info | Does the importance of AI risk depend on caring about transhumanist utopias?
One does not need to care about transhumanist utopias to recognize the importance of AI risk. Misaligned artificial intelligence poses a serious threat to the continued flourishing, and maybe even [continued existence](/?state=89LL&question=W... |
9bfc5b48-9614-4bfd-8e33-d6ab03c19968 | trentmkelly/LessWrong-43k | LessWrong | Why not constrain wetlabs instead of AI?
Most of the object-level stories about how misaligned AI goes wrong involve either nanotechnology or bio-risk or both. Certainly I can (and have, and will again) tell a story about AI x-risk that doesn't involve anything at the molecular level. A sufficient amount of (macroscal... |
d9000fa5-35a3-47c8-918e-b4d955110df6 | trentmkelly/LessWrong-43k | LessWrong | Utopic Nightmares
Utopias are hard to believe in.
Our basis for normality - and thus for our frame of reference for the world - obviously follows from what we’ve seen and experienced in the world around us. But utopias are hard to write, so, never seeing them, we’re left with a glaring emptiness in place of cached... |
af2806c7-5ea1-4a6c-80e4-e0327f68f655 | trentmkelly/LessWrong-43k | LessWrong | AI #70: A Beautiful Sonnet
They said it couldn’t be done.
No, not Claude Sonnet 3.5 becoming the clear best model.
No, not the Claude-Sonnet-empowered automatic meme generators. Those were whipped together in five minutes.
They said I would never get quiet time and catch up. Well, I showed them!
That’s right. Yes,... |
22bfd8ff-c72a-41f8-aba9-369fe86d5923 | trentmkelly/LessWrong-43k | LessWrong | HPMoR.com
Josh's mirror of Harry Potter and the Methods of Rationality has been redesigned by Lightwave (who also did IntelligenceExplosion.com, Friendly-AI.com, and lukeprog.com), and it is now located at a simpler URL: HPMoR.com. Thanks also to Louie who put together this "facelift" project. |
090fa725-c64a-43ed-8731-344e92a4844c | trentmkelly/LessWrong-43k | LessWrong | Scoping LLMs
Emile Delcourt, David Baek, Adriano Hernandez, Erik Nordby with advising from Apart Lab Studio
Introduction & Problem Statement
Helpful, Harmless, and Honest (”HHH”, Askell 2021) is a framework for aligning large language models (LLMs) with human values and expectations. In this context, "helpful" means... |
10e0da6d-703f-44b2-b46e-7e1f366e604c | trentmkelly/LessWrong-43k | LessWrong | Frontline of AGI Alignment
I created a post for the Future Fund AI World Prize (AGI xRisk Competition) only days after seeing it mentioned on twitter, but I don’t really follow people on the alignment-side of the isle, so of course that post fell flat. As I’ve finally decided to make the formal post here according to... |
fd96cc38-599c-4eb7-8173-270d18799891 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Pretraining Language Models with Human Preferences
*This post summarizes the main results from our recently released paper*[*Pretraining Language Models with Human Preferences*](https://arxiv.org/abs/2302.08582)*, and puts them in the broader context of AI safety. For a quick summary of the paper, take a look at*[*our... |
a6e901f8-1657-44a4-bf36-5866114e777d | StampyAI/alignment-research-dataset/lesswrong | LessWrong | They gave LLMs access to physics simulators
Over at Google, large language models have been plugged into physics simulators to help them share a world model with their human interlocutors, resulting in big performance gains. They call it Mind's Eye.
This is how the authors describe the work:
> Correct and compl... |
5d4f9605-f251-458f-8194-ee6122f10cea | StampyAI/alignment-research-dataset/arxiv | Arxiv | COLA: Consistent Learning with Opponent-Learning Awareness.
1 Introduction
---------------
Much research in deep multi-agent reinforcement learning (MARL) has focused on zero-sum games like Starcraft and Go (Silver et al., [2017](#bib.bib34); Vinyals et al., [2019](#bib.bib36)) or fully cooperative settings (Oroojl... |
37855f4a-1d51-42ec-8b3f-d63569c5b4f5 | trentmkelly/LessWrong-43k | LessWrong | Book Review: AI Safety and Security
I just read Roman Yampolskiy's new 500-pages-multi-author anthology: Artificial Intelligence Safety and Security to get a comprehensive view of the field. Here is my attempt at reviewing the book while explaining the main points of those 28 independent chapters.
General Comments a... |
2257e95e-fdff-4423-942f-a81373c8d61c | trentmkelly/LessWrong-43k | LessWrong | Internal Availability
Edit: Following mixed reception, I decided to split this part out of the latest post in my sequence on reinforcement learning. It wasn't clear enough, and anyway didn't belong there.
I'm posting this hopefully better version to Discussion, and welcome further comments on content and style.
-... |
62547037-650d-4a95-a724-822b548594b7 | trentmkelly/LessWrong-43k | LessWrong | History repeats itself: Frontier AI labs acting like Amazon in the early 2000s
(I'm not new to posting on LessWrong. This is a new account I am using to crosspost LW-relevant content from my Substack.)
I recently read "The Everything War", a 2024 book by Dana Mattioli. The thesis of the book is that the Amazon of to... |
44f1f0e2-7740-43ac-b069-6d5f17a5de0d | trentmkelly/LessWrong-43k | LessWrong | Can corrigibility be learned safely?
EDIT: Please note that the way I use the word "corrigibility" in this post isn't quite how Paul uses it. See this thread for clarification.
This is mostly a reply to Paul Christiano's Universality and security amplification and assumes familiarity with that post as well as Paul's ... |
f3551346-8e25-46ec-a52d-29d2f429702b | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Alignment is hard. Communicating that, might be harder
*Note: this is my attempt to articulate why I think it's so difficult to discuss issues concerning AI safety with non-EAs/Rationalists, based on my experience. Thanks to McKenna Fitzgerald for our recent conversation about this, among other topics.*
The curre... |
0aa1f175-604e-45f0-83b0-af3e3cb93b50 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | A central AI alignment problem: capabilities generalization, and the sharp left turn
(*This post was factored out of a larger post that I (Nate Soares) wrote, with help from Rob Bensinger, who also rearranged some pieces and added some text to smooth things out. I'm not terribly happy with it, but am posting it anyway... |
ce2fdea0-20b7-486a-acd9-400cee303999 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Certified Defenses against Adversarial Examples
1 Introduction
---------------
Despite the impressive (and sometimes even superhuman) accuracies of machine learning
on diverse tasks such as object recognition (He et al., [2015](#bib.bib19)), speech recognition
(Xiong et al., [2016](#bib.bib56)), and playing Go (Sil... |
5c8161f5-73d2-4e10-b998-cb8c339312d1 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Updates on FLI's Value Aligment Map?
Around January 2021, I came across the Future of Life Institute's [Value Alignment Map](https://futureoflife.org/valuealignmentmap/), and found it to be a useful resource for getting a sense of the different alignment topics. I was going through it again today, and recalled that I ... |
2ddde1ae-9c00-4821-8b12-b8c34f5f3522 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | What are the arguments that support China building AGI+ if Western companies delay/pause AI development?
In nearly every discussion I've engaged in relating to the potential delay or pause in AI research, multiple people have responded with the quip: "If we don't build AGI, then China will, which is an even worse poss... |
1eff0044-2787-45ee-8b4a-4049df0dd037 | trentmkelly/LessWrong-43k | LessWrong | Study: Encouraging Obedience Considered Harmful
A while back I did a couple of posts on the care and feeding of young rationalists. Though it is not new, I recently found a truly excellent post on this topic, in Dale Mcgowan's blog, The Meming of Life. The post details a survey carried out on ordinary citizens of Hitl... |
e2545a87-e7a9-467a-b506-2ab5a3158856 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Washington, D.C.: Why Ad Hominem Works
Discussion article for the meetup : Washington, D.C.: Why Ad Hominem Works
WHEN: 15 November 2015 03:00:00PM (-0500)
WHERE: Navy Memorial
Note: meetup location has changed! Courtyard is closed today, meetup will happen at the Navy Memorial.
We'll be talking this week... |
98e58c97-7957-4af6-aada-4fc1a6997412 | trentmkelly/LessWrong-43k | LessWrong | On Llama-3 and Dwarkesh Patel’s Podcast with Zuckerberg
It was all quiet. Then it wasn’t.
Note the timestamps on both of these.
Dwarkesh Patel did a podcast with Mark Zuckerberg on the 18th. It was timed to coincide with the release of much of Llama-3, very much the approach of telling your story directly. Dwarkesh ... |
f47931b1-2da1-4f8d-adec-498c7faad39f | trentmkelly/LessWrong-43k | LessWrong | Elon talked with senior Chinese leadership about AI X-risk
As part of his Twitter Spaces with RJK jr. Elon said (Youtube-based transcript):
> On my recent trip to China, with the senior leadership there, we had I think some very productive discussions on artificial intelligence risks and the the need for some oversig... |
5d23a8c9-7449-4780-bf9e-075156291a09 | StampyAI/alignment-research-dataset/blogs | Blogs | More Christiano, Cotra, and Yudkowsky on AI progress
This post is a transcript of a discussion between Paul Christiano, Ajeya Cotra, and Eliezer Yudkowsky (with some comments from Rob Bensinger, Richard Ngo, and Carl Shulman), continuing from [1](https://www.lesswrong.com/posts/vwLxd6hhFvPbvKmBH/yudkowsky-and-christia... |
378f17ad-074d-4ce0-a0e1-f396a833db88 | StampyAI/alignment-research-dataset/arbital | Arbital | Logistic function
The logistic function is a [https://arbital.com/p/-sigmoid](https://arbital.com/p/-sigmoid) function that maps the [real numbers](https://arbital.com/p/4bc) to the unit interval $(0, 1)$ using the formula $\displaystyle f(x) = \frac{1}{1 + e^{-x}}$.
More generally, there exists a [family](https://ar... |
587b595e-e105-46cf-a76a-ac3eec0d61ec | trentmkelly/LessWrong-43k | LessWrong | Desperation hamster wheels
In my first few jobs, I felt desperate to have an impact. I was often filled with anxiety that I might not be able to. My soul ached. Fear propelled me to action. I remember sitting in a coffee shop one Saturday trying to read a book that I thought would help me learn about biosafety, an imp... |
32864608-d151-4bc2-991a-261ff5ed3495 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Philadelphia LW: Macroeconomics crash course and general meetup
Discussion article for the meetup : Philadelphia LW: Macroeconomics crash course and general meetup
WHEN: 15 February 2012 06:30:00PM (-0500)
WHERE: 1301 Chestnut Street Philadelphia, PA 19107-3521 US
We'll be meeting at the Starbucks in Macy'... |
a88bc847-d18d-4ce1-86df-3de87343a907 | trentmkelly/LessWrong-43k | LessWrong | Software Engineering: Getting Hired and Promoted
I'm continuing my series of posts this week to earn more GHT for my donation lottery with another post about my job. This time, I'm coming from the angle of telling you what I tell other software engineers about getting hired and getting promoted.
So, you want to get h... |
0e65f4b5-be98-470a-b445-1ad7942ff58d | trentmkelly/LessWrong-43k | LessWrong | [Linkpost] Existential Risk Analysis in Empirical Research Papers
I’ve noted before that it can be difficult to separate progress in safety from progress in capabilities. However, doing so is important, as we want to ensure that we are making differential progress on safety, rather than just advancing safety as a cons... |
10b56436-7bea-4723-9e60-4e57702f6ee2 | trentmkelly/LessWrong-43k | LessWrong | Using axis lines for good or evil
Say you want to plot some data. You could just plot it by itself:
Or you could put lines on the left and bottom:
Or you could put lines everywhere:
Or you could be weird:
Which is right? Many people treat this as an aesthetic choice. But I’d like to suggest an unambiguous ... |
00b451ba-662f-4208-8091-3fe530214928 | trentmkelly/LessWrong-43k | LessWrong | Deconditioning Aversion to Dislike
It doesn't matter who dislikes me. It doesn't matter who hates me. It doesn't even matter who likes me. All that matters is how many people love me because I live under the rule of law in a city where the people I have the potential to interact with vastly outnumber the people I have... |
fdf41c8e-222c-454b-9d87-9d7936af4777 | trentmkelly/LessWrong-43k | LessWrong | Unlimited Pomodoro Works: My Scheduling System
Related: The Power of Pomodoros, Working Hurts Less Than Procrastinating, Cached Procrastination
Follow-up To: Reinforcement and Short-Term Rewards as Anti-Akratic
I'm still working on cleaning up my scheduling system for release, like I mentioned in the comments to my ... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.