id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
cc816be6-d1f4-452b-af02-3586437126d8
trentmkelly/LessWrong-43k
LessWrong
marine cloud brightening Various geoengineering schemes have been proposed to mitigate global warming. Some prominent schemes I don't like are accelerated weathering and stratospheric aerosol injection. I think marine cloud brightening is a better proposal than those. accelerated weathering To potentially absorb 1 t...
80736817-f6f6-45bf-9c2e-4bfb441a5f4f
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Artificial Sandwiching: When can we test scalable alignment protocols without humans? *Epistemic status: Not a fleshed-out proposal. Brainstorming/eliciting ideas.* *Thanks to Ben Mann,  Pablo Moreno, and Jared Kaplan for feedback on early drafts.* Overview -------- * I’m convinced [sandwiching](https://www.alignme...
3e9d3d0f-b7e1-4771-b505-425a9de6a08a
trentmkelly/LessWrong-43k
LessWrong
Meetup : LW-cologne Meetup Discussion article for the meetup : LW-cologne Meetup WHEN: 13 June 2015 05:00:00PM (+0200) WHERE: Mareinwe 43, 50858 Köln Hi everyone, time for our monthly meetup again. I decided not to change the date, since there seemed to be mixed preferences for June (Martin 2nd weekend, Mimi 1st w...
e2e4aa3e-660c-4501-afdb-704641080ce1
trentmkelly/LessWrong-43k
LessWrong
Weekly LW Meetups: Fort Collins and Melbourne There are upcoming irregularly scheduled Less Wrong meetups in: * Fort Collins, Colorado Meetup: 04 January 2012 07:00PM * First Philadelphia Meetup of 2012: 08 January 2012 02:00PM * [TENTATIVE] Portland Meetup?: 14 January 2012 12:00PM * San Diego experimental meetu...
ad4b0430-c155-47d0-9f8d-d2a4ca63d09e
trentmkelly/LessWrong-43k
LessWrong
Towards optimal play as Villager in a mixed game On Twitter, Freyja wrote: > Things capitalism is trash at: > Valuing preferences of anything other than adults who earn money (i.e. future people, non-humans) > Pricing non-standardisable goods (i.e. information) > Playing nicely with non-quantifiable values + objectiv...
e01c6f5d-685a-40e2-a5c8-eeb363f5752d
trentmkelly/LessWrong-43k
LessWrong
Synthesising divergent preferences: an example in population ethics This post is an example of how one could go about sythesising contradictory human preferences and meta-preferences, using a somewhat simplified version of the method of this post. The aim is to illustrate how I imagine the process going. I'll take th...
51278208-be94-4c63-a172-4d015de8b8f0
trentmkelly/LessWrong-43k
LessWrong
What are good rationality exercises? I want to know what are good rationality exercises. I was just on a call with Liron and PhilH, hanging out after the weekly LessWrong weekend event, and we discussed exercises that could happen on LessWrong. Here is the list we generated: * Thinking Physics * Fermi Estimates ...
9410b713-c412-4528-b95d-c588f1a895be
trentmkelly/LessWrong-43k
LessWrong
What should rationalists think about the recent claims that air force pilots observed UFOs? NY times has a new material about possible observations of UFOs by air force pilots. Should rationalists use their arsenal of new ideas about how the world works – simulation, superintelligent AIs etc – to explain these things,...
8462d739-aaac-4fe0-bec9-7148d7bffcb0
trentmkelly/LessWrong-43k
LessWrong
Possible personal implications of the Israeli Hunger-Probation study I'll let Psychohistorian provide the set-up, in case the reference is unfamiliar. Anyway, my wife's expecting, and that means many, many tests, ultrasounds, etc. that she ends up going through. She had an ultrasound done yesterday that was a follow-u...
a9bd09e2-80aa-49a0-a970-a1ffaea00624
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
Towards AI You Can Rely On (Aleksander Madry, Foundations of Safe Learning Workshop) we to be nice to the environment we haven't printed physical programs but you can see the list of speakers in the schedule that workshop safe learning a I we have a really fantastic lineup of speakers today we're very fortunate to be ...
4c160b92-0c88-4854-8d8e-9d703e8667e7
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Worse Than Random **Previously in series**:  [Lawful Uncertainty](/lw/vo/lawful_uncertainty/) You may have noticed a certain trend in recent posts:  I've been arguing that randomness hath no power, that there is no beauty in entropy, nor yet strength from noise. If one were to formalize the argument, it would prob...
eb1f9b73-bf2b-4bd0-8b50-e7680c2d9b9b
trentmkelly/LessWrong-43k
LessWrong
Welcome to ACX/Less Wrong Budapest Hello, we are your sort of nerds in Budapest! Come and join us. Usually we have a discussion meetup once a month, where there is some set of articles that we read and then talk about more deeply. If there are enough people we specifically will break up into smaller groups.  Honestl...
1f0a85e9-2de8-4eee-b227-6674d6412f4a
trentmkelly/LessWrong-43k
LessWrong
Prospects for Alignment Automation: Interpretability Case Study For human-level AI (HLAI) we will need robust control or alignment methods. Assuming short timelines to HLAI, the tractability of automating safety research becomes central. In this post, I will make the case that safety-relevant progress on automated int...
6e5fd131-aafc-47a0-a3a3-21c39430a23a
StampyAI/alignment-research-dataset/arxiv
Arxiv
Recursively Summarizing Books with Human Feedback ### 1 Introduction To train an ML model on a new task, we need a training signal that tells the model which behaviors are better and which are worse. For some tasks, like playing a video game, this training signal can be calculated automatically. However, for many ...
364f411e-2527-430c-b607-910abd1328b1
StampyAI/alignment-research-dataset/arxiv
Arxiv
Forecasting AI Progress: A Research Agenda Forecasting AI Progress: A Research Agenda Ross Gruetzemacher*, Florian Dorner ​ † ​ , Niko Bernaola-Alvarez ​ ‡ ​ , Charlie Giattino ​ § ​ , David Manheim ​ ** *Auburn University, ​ † ​ Freie Universität Berlin, ​ ‡ ​ Universidad Politécnica de Madrid, ​ § ​ U...
dd9bbb85-de82-4dc3-b36c-fef4d271dafb
trentmkelly/LessWrong-43k
LessWrong
Meetup : Moscow: Applied Rationality Discussion article for the meetup : Moscow: Applied Rationality WHEN: 19 January 2013 04:00:00PM (+0400) WHERE: Russia, Moscow, ulitsa L'va Tolstogo 16 Our venue is changed: We will meet in the Yandex Money office, this is the second door at the address I mentioned. Please use ...
73ba20a0-de97-4535-b97b-07b6ece827f6
trentmkelly/LessWrong-43k
LessWrong
AtP*: An efficient and scalable method for localizing LLM behaviour to components Authors: János Kramár, Tom Lieberum, Rohin Shah, Neel Nanda A new paper from the Google DeepMind mechanistic interpretability team, from core contributors János Kramár and Tom Lieberum Tweet thread summary, paper Abstract: > Activati...
88ddf3c5-7113-4b2c-b3b5-35794dbae335
StampyAI/alignment-research-dataset/blogs
Blogs
Beyond fire alarms: freeing the groupstruck *Katja Grace, 9 September 2021* *[Content warning: death in fires, death in machine apocalypse]* ‘No fire alarms for AGI’ ------------------------ Eliezer Yudkowsky wrote that ‘[there’s no fire alarm for Artificial General Intelligence](https://intelligence.org/2017/10...
841f476c-8f2e-428a-a6c2-e4fea03e4986
trentmkelly/LessWrong-43k
LessWrong
September 2020 gwern.net newsletter None
e9a87f3d-e4d6-4fc1-98fc-c3ee4469c399
trentmkelly/LessWrong-43k
LessWrong
The principle of no non-Apologies Original on my website TL;DR: Principle of no non-Apologies: “Distinguish between saying I’m sorry and apologizing. Don’t give non-Apologies.” Do not Apologize when you don’t agree that you fucked up. When you fucked up, own the fuck-up and, if it’s systematic, commit to reducing fut...
752e6154-4184-4239-844f-f12bff9b4d74
trentmkelly/LessWrong-43k
LessWrong
Integrating the Lindy Effect Suppose the following: 1. Your intelligence is directly proportional to how many useful things you know. 2. Your intelligence increases when your learn things and decreases as the world changes and the things you know go out-of-date. How quickly the things you know become irrelevant is ...
9ccf106f-22ff-4c09-9a71-7bfe175c9079
trentmkelly/LessWrong-43k
LessWrong
Pomodoro for Programmers Unless you’ve been living under a productivity rock, you probably have heard of the Pomodoro Technique, where you use a timer to do 25 minutes of focused work, and then take a five minute break. I used to use this technique a lot, up until I started doing computer programming. You see, with ...
6f2a404b-9b6e-4c9f-bc1d-3685107fe8ec
trentmkelly/LessWrong-43k
LessWrong
Open thread, 11-17 March 2014 If it's worth saying, but not worth its own post (even in Discussion), then it goes here.
0b53080d-1d35-4281-ac93-978320b293cd
trentmkelly/LessWrong-43k
LessWrong
Comparative Advantage is Not About Trade Braudel is probably the most impressive historian I have read. His quantitative estimates of premodern populations and crop yields are exactly the sort of foundation you’d think any understanding of history would be based upon. Yet reading his magnum opus, it became steadily cl...
d9549122-2733-4579-97ae-beb8ec5e9a14
trentmkelly/LessWrong-43k
LessWrong
What is Interpretability? In this post we lay out some ideas around framing interpretability research which we have found quite useful. Our framing is goal-oriented, which we believe is important for making sure interpretability research is meaningful. We also go over a variety of dimensions which we think are useful ...
048b5fa8-94bf-471b-9352-5e51593f918e
trentmkelly/LessWrong-43k
LessWrong
Transformative trustbuilding via advancements in decentralized lie detection Although the emergence of functional lie detection would be an obvious total paradigm shift for the entire court system, the author didn’t seem to realize that this is also an obvious total paradigm shift for much bigger things, e.g. hiring, ...
01c77165-7bb4-45a1-84dc-c36c30c6cc19
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Is AI like disk drives? One finding of an extensive theoretical and empirical economics literature on technology races is that firms that trail the current leader might tend to innovate more ([Reinganum 1983](https://www.jstor.org/stable/1816571); [Lerner 1997](https://www.jstor.org/stable/2555803)). My question is ab...
04ccae03-ee82-46c6-b91e-000adecd92a4
trentmkelly/LessWrong-43k
LessWrong
[RESEARCH] Marijuana may prevent Alzheimer's disease In a post last August called Alzheimer's vs. Cryonics, I left the following comment: > 23andMe recently showed that people of my genotype are more than twice as likely to develop Alzheimer's as others in my ethnic group. I have a 15% chance of getting Alzheimer's b...
1d95e99c-ebdb-4d2c-9983-5b88a402fa00
trentmkelly/LessWrong-43k
LessWrong
Meetup : Fort Collins, Colorado Meetup Discussion article for the meetup : Fort Collins, Colorado Meetup WHEN: 22 February 2012 07:00:00PM (-0700) WHERE: 144 North College Avenue, Fort Collins, CO 80524 We have a guest from London/Budapest visiting this week. Come discuss intelligence augmentation tools. Discussi...
5ea6a616-11da-4ef5-8eda-4406f66605b6
trentmkelly/LessWrong-43k
LessWrong
[Link] Less Wrong and Agency in the Huffington Post I'm guessing this is the first time that links to Less Wrong appeared in a Huffington Post article, and three links at that! Please correct me if I'm wrong on this one. I'm also guessing this is the first time that the concept of agency in the rationalist sense was d...
7d40c3ff-49b6-4179-80bc-56ac53469ce5
trentmkelly/LessWrong-43k
LessWrong
Counting Objections to Housing Over the past six months there's been a huge amount of discussion in the Davis Square Facebook group about a proposal to build a 25-story building in Davis Square: retail on the ground floor, 500 units of housing above, 100 of the units affordable. I wrote about this a few weeks ago, wei...
eaa26037-f1d2-437b-85b1-4e60be380731
trentmkelly/LessWrong-43k
LessWrong
What Is Childhood Supposed To Be? I don’t have children. I’d like to, one day, and I think a lot about what their childhoods might be like, especially as compared to mine. I look at reports of teen mental health, at college admission rates and tuition levels. I talk to friends of mine who are teachers and involved w...
307ce276-1b02-4991-a846-9b88ef67f798
trentmkelly/LessWrong-43k
LessWrong
July 5: Anders Sandberg on Information Hazards (SSC Online Meetup) Dr Anders Sandberg, Senior Research Fellow at the Future of Humanity Institute, will speak on "Information Hazards: How do we think about (and handle) risky information." We will also allow plenty of time for online mingling and discussion. Click he...
6c9e1995-de89-49b6-a8c1-576406c6e14d
trentmkelly/LessWrong-43k
LessWrong
Open Challenges in Representation Engineering This post summarizes the taxonomy, challenges, and opportunities from a survey paper on Representation Engineering that we’ve written with Sahar Abdelnabi, David Krueger, and Mario Fritz. If you’re familiar with RepE feel free to skip to the “Challenges” and “Opportunitie...
446f387e-0038-4c59-93ac-673ae5e17324
trentmkelly/LessWrong-43k
LessWrong
How do you concisely communicate & navigate the politics / culture at your job working at a large corporation or institution? Previously I've worked at small companies and state government or educational institutions. Now I work at a ~30,000 person private corporation and need to better adapt to that cultural situatio...
57a5b12d-25bd-4989-bf79-e4fc4b1ba05f
trentmkelly/LessWrong-43k
LessWrong
The Convergent Path to the Stars Crossposted on the EA Forum. This document is a shallow review of selection pressures and filters influencing the characteristics of the Shapers of Space-Faring Civilizations (SFCs). SFC Shapers are the agents within Intelligent Civilizations influencing the most the SFCs they will cr...
7803705b-4b9c-41b2-ae2f-d61a70135797
StampyAI/alignment-research-dataset/arxiv
Arxiv
AutoML-Zero: Evolving Machine Learning Algorithms From Scratch 1 Introduction --------------- In recent years, neural networks have reached remarkable performance on key tasks and seen a fast increase in their popularity [e.g. He et al., [2015](#bib.bib32); Silver et al., [2016](#bib.bib80); Wu et al., [2016](#bib....
0bf7d673-cbfc-4cc4-9c82-6c3c7255bffe
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
AI & Policy 1/3: On knowing the effect of today’s policies on Transformative AI risks, and the case for institutional improvements. **Introduction** ================ **The question: What policies will impact AI risks?** ----------------------------------------------------- This post is looking to answer the challeng...
666f8685-13e5-48fd-b341-78b5c5d012a6
trentmkelly/LessWrong-43k
LessWrong
Meetup : Berkeley: Board games Discussion article for the meetup : Berkeley: Board games WHEN: 30 January 2013 07:30:00PM (-0800) WHERE: Berkeley, CA Hello all! This week's meetup will be a board games night. Come meet people and play a board game from Zendo's library, or bring a game that you want to play. I will ...
d995863d-df3d-48f9-bc03-5cc796b9ab6e
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Simpler explanations of AGI risk We're getting a shot at presenting our concerns about AI X-risk to the general public. It would be useful to have a brief presentation that plays well with less-technical people, or technical people who don't want to listen to a half hour of explanation just at that moment. The other g...
8358f5fa-db6f-4327-89dc-dbe6c6d1e2e5
trentmkelly/LessWrong-43k
LessWrong
British and American Connotations As an American who works with some people who speak British English, the language differences are usually not a problem. Most words mean the same thing, and those that don't are usually concrete enough not to cause confusion (ex: lift, flat, chips). The tricky ones, though, are the on...
55b2518c-48f6-4e67-9e73-4cbfef5d289e
StampyAI/alignment-research-dataset/agisf
AGI Safety Fund
Machine Learning for Humans, Part 2.1: Supervised Learning The two tasks of supervised learning: regression and classification. Linear regression, loss functions, and gradient descent. ----------------------------------------------------------------------------------------------------------------------------- [![Vish...
b845447d-f991-4e92-aa16-0e1d617ccedc
trentmkelly/LessWrong-43k
LessWrong
Meetup : Salt Lake City LW Meetup #2 Logistics Discussion article for the meetup : Salt Lake City LW Meetup #2 Logistics WHEN: 04 February 2012 03:00:00PM (-0700) WHERE: Downtown central Salt Lake City (near the library) Update: see lesswrong.com/meetups for info on the next Meetup Note: this announcement is to ga...
2493608f-99f1-464c-b9c8-cf985e7a5d65
trentmkelly/LessWrong-43k
LessWrong
How well does your research adress the theory-practice gap? (5-6 min read time without exercise instructions) Motivation for post: Finding a way to map mathematics & areas of academic research to the alignment problem seems to be a core part of theoretical alignment and agent foundations work. Knowing how to frame di...
70ad9e73-2087-4a4b-bbba-323581a9b887
trentmkelly/LessWrong-43k
LessWrong
Testing ChatGPT 3.5 for political biases using roleplaying prompts I wanted to test ChatGPT for political biases, using roleplaying prompts to have the chatbot roleplay as either a Democrat, or a Republican. The results were…interesting. Tl;dr: Yes, ChatGPT is biased, up to the point where it actually misrepresents th...
24b25320-abc5-47d4-a010-1fb7e0a1cfd1
trentmkelly/LessWrong-43k
LessWrong
Timeless Beauty Followup to:  Timeless Physics One of the great surprises of humanity's early study of physics was that there were universal laws, that the heavens were governed by the same order as the Earth:  Laws that hold in all times, in all places, without known exception. Sometimes we discover a seeming except...
7c880022-ffa4-4d98-b4cc-15f407e7b41c
trentmkelly/LessWrong-43k
LessWrong
Value learning subproblem: learning goals of simple agents A potential problem with inverse reinforcement learning as a way for an AI to learn human values is that human actions might not contain enough information to infer human values from accurately enough. If this is the case, then it might be necessary to figure ...
12a93de2-8f13-438c-9c0b-f592a17dfc89
trentmkelly/LessWrong-43k
LessWrong
Resetting my perception of something The guy who taught me how to eat sushi a couple months ago explained that I should get a slice of ginger before trying a different kind of sushi in order to "reset my taste buds" (probably broscience) and feel the taste anew (it works). That's also the exact problem that programmer...
434bb00e-4392-46b6-84f6-8ed6e9e021c0
trentmkelly/LessWrong-43k
LessWrong
Chapter 39: Pretending to be Wise, Pt 1 Whistle. Tick. Bzzzt. Ding. Glorp. Pop. Splat. Chime. Toot. Puff. Tinkle. Bubble. Beep. Thud. Crackle. Whoosh. Hiss. Pffft. Whirr. Professor Flitwick had silently passed Harry a folded parchment during Charms class that Monday, and the note had said that Harry was to visit the ...
f01c9fca-1186-49b9-8428-5da72740019d
trentmkelly/LessWrong-43k
LessWrong
Review of Kurzweil, 'The Singularity is Near' Ray Kurzweil's writings are the best-known expression of Singularity memes, so I figured it's about time I read his 2005 best-seller The Singularity is Near. Though earlier users of the term "technological Singularity" used it to refer to the arrival of machine superintel...
d6b0cf0a-b5ab-4263-b27e-97bf7adf8dcc
StampyAI/alignment-research-dataset/arxiv
Arxiv
Pretrained Transformers as Universal Computation Engines 1 Introduction --------------- The transformer architecture (Vaswani et al., [2017](#bib.bib1 "Attention is all you need")) has shown broad successes in deep learning, serving as the backbone of large models for tasks such as modeling natural language (Brown ...
bb49cb2e-93c1-47d5-bd9e-78617b9d9880
trentmkelly/LessWrong-43k
LessWrong
Should we postpone AGI until we reach safety? Should we postpone AGI, until its risks have fallen below a certain level, thereby applying the precautionary principle? And if so, would setting up policy be a promising way to achieve this? As I argued here in the comments, I think calling for precautionary principle-po...
7cfab2e3-4f51-40d7-b6c7-309ad329515a
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Counterfactual Reprogramming Decision Theory Surprisingly, Ben Goertzel's ([2010](http://goertzel.org/CounterfactualReprogrammingDecisionTheory.pdf)) Counterfactual Reprogramming Decision Theory (CRDT) has not been discussed even once on Less Wrong, so I present this discussion post as an opportunity to do so. Here ...
daf29dad-359c-4799-b778-cb0bc556c23a
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post2005 This is an ELK proposal that received a prize in the category " Train a reporter that is useful to an auxiliary AI ". It was coauthored with Vikrant Varma and Ramana Kumar and advised by Rohin Shah. Training strategy and why it might work We use an architecture with a knowledge bottleneck to train the mod...
9ce7b0da-c061-4451-81cb-5f47295153bb
StampyAI/alignment-research-dataset/lesswrong
LessWrong
The Compleat Cybornaut A [cluster](https://www.lesswrong.com/tag/simulator-theory) of conceptual frameworks and research programmes have coalesced around a 2022 post by [janus](https://www.lesswrong.com/posts/vJFdjigzmcXMhNTsx/simulators), which introduced language models as ‘simulators’ (of other types of AIs such as...
e7102c85-3045-4318-b192-c3b8d4d154fa
trentmkelly/LessWrong-43k
LessWrong
Proof-of-Concept Debugger for a Small LLM We want to show that it is possible to build an LLM “debugger” using SAE features and have developed a prototype that automates circuit visualizations for arbitrary prompts. With a few improvements to existing techniques (notably, “cluster resampling”, which is a form of activ...
6b66758d-cc0e-4bfe-af93-a2d4ec7a4b75
StampyAI/alignment-research-dataset/arxiv
Arxiv
Thinking Fast and Slow with Deep Learning and Tree Search 1 Introduction --------------- According to dual-process theory[[1](#bib.bib1), [2](#bib.bib2)], human reasoning consists of two different kinds of thinking. System 1, is a fast, unconscious and automatic mode of thought, also known as intuition or heuristic...
b128d7f9-9fd2-4cd8-833d-1f27a79d964b
trentmkelly/LessWrong-43k
LessWrong
On Emergent Misalignment One hell of a paper dropped this week. It turns out that if you fine-tune models, especially GPT-4o and Qwen2.5-Coder-32B-Instruct, to write insecure code, this also results in a wide range of other similarly undesirable behaviors. They more or less grow a mustache and become their evil twin....
1e00433b-b548-4563-a66b-eb4e280d729d
trentmkelly/LessWrong-43k
LessWrong
How Can Average People Contribute to AI Safety? Introduction By now you've probably read about how AI and AGI could have a transformative effect on the future and how AGI could even be an existential risk. But if you're worried about AI risk and not an AI researcher or policymaker, can you really do anything about it ...
5ad1926f-591e-4a8f-82fc-b2fd647acc21
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Tolerate Tolerance Today's post, Tolerate Tolerance was originally published on 21 March 2009. A summary (taken from the LW wiki):   > One of the likely characteristics of someone who sets out to be a "rationalist" is a lower-than-usual tolerance for flawed thinking. This makes it very important to toler...
d76d84d8-cc92-4a0a-8a18-a64e57843a2f
trentmkelly/LessWrong-43k
LessWrong
How likely is our view of the cosmos? In terms of what we see in the night sky, are we a statistical anomaly compared to the average star system? That is, what percentage of stars with potentially habitable planets exist close enough to other stars to see a significant number of them in the night sky, have a mostly un...
9e543769-67cd-4c96-b182-af35be2fdbda
trentmkelly/LessWrong-43k
LessWrong
[Link] Space Stasis: What the strange persistence of rockets can teach us about innovation http://www.slate.com/id/2283469/pagenum/all/ It's a long article, but the most relevant stuff is at the end, about how we're pretty much locked into the existing rocket technologies: > That is not, however, the most important ...
ba8601c9-8bff-4e5d-8453-abd77dbb4a84
trentmkelly/LessWrong-43k
LessWrong
Organizing the FOOM debate Why was the FOOM debate not conducted using any debate tools? Perhaps that would enable agreement theorems to actually work and the debate to be resolved. Would someone want to go through the debate and try to organize it?
cb5d788c-b521-4278-83ec-2cc8d40625c6
trentmkelly/LessWrong-43k
LessWrong
What rationality material should I teach in my game theory course I will soon be teaching a 50 person game theory class at Smith College and I want to include material on how to be rational. What do you think I should teach/assign?  The course has a one semester calculus requirement. Here is material I'm considering: ...
012e1ab4-8247-4a63-8e12-53450f4f2b10
trentmkelly/LessWrong-43k
LessWrong
[LINK] Report on the Fourth Conference on Artificial General Intelligence http://hplusmagazine.com/2011/09/01/report-on-the-fourth-conference-on-artificial-general-intelligence/ > By: Ben Goertzel > Published: September 1, 2011 > The Fourth Conference on Artificial General Intelligence (AGI-11) was held on Google’s ...
7d357263-6591-45c4-b395-fd9a98d4d9e7
StampyAI/alignment-research-dataset/arxiv
Arxiv
DeepLine: AutoML Tool for Pipelines Generation using Deep Reinforcement Learning and Hierarchical Actions Filtering Introduction ------------ The explosion of digital data has made the use of machine learning (ML) more ubiquitous than ever before. Machine learning is now applied to almost any aspect of organization...
4f0c2d84-fd75-47b5-8fd5-c832eb36cd60
trentmkelly/LessWrong-43k
LessWrong
Scaling and evaluating sparse autoencoders [Blog] [Paper] [Visualizer] Abstract: > Sparse autoencoders provide a promising unsupervised approach for extracting interpretable features from a language model by reconstructing activations from a sparse bottleneck layer.  Since language models learn many concepts, autoen...
520d3c40-266f-481e-b542-723669159395
trentmkelly/LessWrong-43k
LessWrong
The Art of the Overbet Previously: The Kelly Criterion Last time I said, never go full Kelly. In practice, I strongly agree with this. Either one should go far over full Kelly because the core Kelly assumptions have broken down and you want to throw the rules out the window, or you are trying to responsibly grow a b...
4f48c536-3388-4c55-8645-c3ca825be15c
trentmkelly/LessWrong-43k
LessWrong
I had a chat with GPT-4 on the future of AI and AI safety Today I gave OpenAI some money to get access to GPT-4. After testing a few examples of reasoning that failed badly in GPT-3, I decided to have a chat with it about the potential and risks of future AI. This is what followed, un-edited with the exception of remo...
e018fa4a-8653-4ac4-98a9-aab2909f351c
trentmkelly/LessWrong-43k
LessWrong
Preparing for ambition When I was in college, I decided that I wanted to spend my life starting startups. It seemed like the perfect "career path". If things really work out for you, the reward is billions of dollars. Awesome. But here's the really cool part: even if you "fail", like that Seinfeld episode, things will...
b7cd9410-c016-488b-8333-e291f1d63f1b
trentmkelly/LessWrong-43k
LessWrong
Chatbots as a Publication Format Overview: Recent developments in AI will change the world in all sorts of ways. It is likely to revolutionize academic research. This is one proposal regarding how AI might be used to improve the way that we communicate arguments (such as in philosophy). I stand by behind the thought t...
0b5feb52-6c53-4881-9989-e27362b12635
trentmkelly/LessWrong-43k
LessWrong
What if our Galaxy isn't full of AI because we're in a Neutral Zone between them? No one can build Von Neuman probes, because for that you likely have AI... and AI built is neutralized as soon as it reaches a certain point. No competition. And they compete because, energy is at a premium in a entropic universe.
a7bee9e5-8181-4d3f-9d00-5021918d71ce
trentmkelly/LessWrong-43k
LessWrong
Cryonics? Should I sign up for cryonics ASAP? Background: 23 y/o male in NYC. No known pre-existing health conditions. Mild → moderate interest in cryonics; in a world without ncov, etc. would probably have a 30-60% chance of signing up.
6b47e393-2b14-4946-a709-6682d30b12f1
StampyAI/alignment-research-dataset/arbital
Arbital
Isomorphism: Intro (Math 0) If two things are essentially the same from a certain perspective, and they only differ in unimportant details, then they are **isomorphic**. ##Comparing Amounts## Consider the Count von Count. He cares only about counting things. He doesn't care what they are, just how many there are. He ...
33921ad5-42fd-4d6f-b938-a2835f2af4a3
trentmkelly/LessWrong-43k
LessWrong
[LINK] Subculture roles (by Brad Hicks) By what looks like popular demand on the Society for Creative Anachronism thread, I'm linking a three-part account of three roles that supposedly exist within every lasting subculture or "volunteer-dominated" organization. The author has no proper authority at all but has person...
9130d993-1dad-4183-aa7f-51005df22d19
trentmkelly/LessWrong-43k
LessWrong
Intuitions and utilitarianism Bryan Caplan: > When backed into a corner, most hard-line utilitarians concede that the standard counter-examples seem extremely persuasive.  They know they’re supposed to think that pushing one fat man in front of a trolley to save five skinny kids ismorally obligatory.  But the opposit...
6e2571f4-e850-4d5c-b29d-b076177a39d8
trentmkelly/LessWrong-43k
LessWrong
Lateral Thinking (AI safety HPMOR fanfic) (Source material) "Now leave your books and loose items at your desks – they will be safe, the screens will watch over them for you – and come down onto this platform. It's time to play a game called Who's the Most Promising Student in the Classroom." -----------------------...
f8a9d5ad-96c0-46d6-822f-1b48e1b71a42
trentmkelly/LessWrong-43k
LessWrong
You've Been Exposed to COVID-19: What Do You Wish You Knew? A thing I've been really frustrated by with a lot of coronavirus advice posts is lack of specificity. What does "quarantine for 14 days" mean? Do I need to poop in a bucket? Can I use my private yard? Let deliveries be placed on my doorstep? If I'm recovered ...
bdb98c08-6393-4ea3-9da8-7683681a9e48
trentmkelly/LessWrong-43k
LessWrong
How-to Transformer Mechanistic Interpretability—in 50 lines of code or less! Produced as part of the SERI ML Alignment Theory Scholars Program - Winter 2022 Cohort. What if I told you that in just one weekend you can get up to speed doing practical Mechanistic Interpretability research on Transformers? Surprised? The...
7c6187ce-820b-43b7-bc3f-75c1caff81bf
trentmkelly/LessWrong-43k
LessWrong
Solstice and Megameetup Preparations for 2013 I'm officially spinning the Solstice and related ritual stuff into something distinct from Less Wrong (there are good reasons to leave LW focusing on straight-up rationality, and I think it should cater more towards "serious business intellectuals" than trying to appeal to...
10caa963-3e6b-4898-a5c2-12c5a25e9aba
trentmkelly/LessWrong-43k
LessWrong
Notes on Phronesis This post examines the virtue of phronesis (“prudence”). It is meant mostly as an exploration of what other people have learned about this virtue, rather than as me expressing my own opinions about it, though I’ve been selective about what I found interesting or credible, according to my own inclina...
2119acb7-15cb-4eea-81b4-57d0e8a7b65a
trentmkelly/LessWrong-43k
LessWrong
Meetup : San Jose Meetup: Park Day (VI) Discussion article for the meetup : San Jose Meetup: Park Day (VI) WHEN: 31 July 2016 03:00:00PM (-0700) WHERE: 438 Coleman Ave, San Jose, CA 95110 You're invited to bring whatever you'd like to see at Park Day - balls, frisbees, food, etc. (Also, consider taking measures aga...
b78f5cfb-9286-4133-bb83-57f25b21aacf
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Announcing aisafety.training *Crossposted to* [*EA Forum*](https://forum.effectivealtruism.org/posts/zWscvNhd3xeGJabEn/announcing-aisafety-training) To help people find what to apply to, [aisafety.training](https://aisafety.training) acts as **a well-maintained living document of AI safety programs, conferences, ...
1bdd626b-27df-48f0-bad8-30ae7be4ca16
trentmkelly/LessWrong-43k
LessWrong
Dagger of Detect Evil A philosopher walked into a magic shop. "Ugh," said Phil. The philosopher leaned his spear against the hatstand and slumped onto a fluffy cushioned chair. "Can I get you something to drink? How about Essence of Essence?" said Wiz, the shop owner. She handed Phil a bottle of perfectly invisible ...
b977bca5-4495-42b8-8480-1bfa9555d20f
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Multi-agent safety *Note: this post is most explicitly about safety in multi-agent training regimes. However, many of the arguments I make are also more broadly applicable - for example, when training a single agent in a complex environment, challenges arising from the environment could play an analogous role to chall...
b16e85f1-b069-4256-af1f-5c260f1f0e39
trentmkelly/LessWrong-43k
LessWrong
Prioritise Tasks by Rating not Sorting I often get paralysed by indecision when I have a bunch of tasks to do - it’s hard to figure out where to start! My standard solution is sorting the tasks into priority order - usually implicitly with a comparison-based sorting method where I go through the list one-by-one and pu...
e70d911e-34cd-4858-b8c5-e6e9104e6a9c
trentmkelly/LessWrong-43k
LessWrong
Rationality Quotes 17 "We take almost all of the decisive steps in our lives as a result of slight inner adjustments of which we are barely conscious."         -- Austerlitz "In both poker and life, you can't read people any better than they can read themselves. You can, if you’re good, very accurately determine if t...
aa74cbdd-76c2-471c-bc26-8c0304ff7625
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"“I just solved the trolley problem.... See, the trolley problem forces you to choose between two versions of letting other people die, but the actual solution is very simple. You sacrifice yourself.” - The Good Place High school: Naive morality When I was a teenager, I had a very simple, naive view of morality. I thou...
05367317-0f13-464a-88ea-aa3e9c9a277f
trentmkelly/LessWrong-43k
LessWrong
Outreach opportunity Ars Technica are holding a competition for people to make a science video up to 3 minutes long "to explain a scientific concept in terms that a high school science class would not only understand, but actually be interested in watching". Prizes in three categories: biology, physics, and mathematic...
efe1d189-b64c-4047-a9c0-1bdc22813f56
trentmkelly/LessWrong-43k
LessWrong
A way to make solving alignment 10.000 times easier. The shorter case for a massive open source simbox project. When you have the code that might be an AGI and probably misaligned, how do you test it in a way that is safe? The world has not yet converged on a way to do this. This post attempts to provide an outline of...
a83c355e-afc9-441b-8318-2d9d7333c04e
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
The Metaethics and Normative Ethics of AGI Value Alignment: Many Questions, Some Implications Short Summary ============= [This](https://drive.google.com/file/d/1fDP0roq_I2sDJY4sq02Y4rnYgcFKwbjj/view) (<--) is a commented research agenda on the metaethics and normative ethics of AGI (artificial general intelligence) ...
b2b2186e-be6f-4157-946b-d9aa87c8916f
trentmkelly/LessWrong-43k
LessWrong
Post-COVID Integration Rituals I'm writing this because I am realizing we're at the cusp of the end of a very long "retreat."  At MAPLE (Monastic Academy), we go on retreat once a month, and so we exit retreat once a month.  This can be quite a shock to the system, so we deliberately set aside a day for integration ...
23bd9e53-a767-4c09-ad7f-ca2628aed885
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Brain-like AGI project "aintelope" Steven Byrnes called for a [brain-like AGI research agenda](https://www.lesswrong.com/s/HzcM2dkCq7fwXBej8/p/tj8AC3vhTnBywdZoA), and three guys from the Hamburg EA community listened. We are excited about Steven's five-star program [15.2.1.2 "Reverse-engineer human social instincts,"...
4e317e12-f0d3-4112-b70e-9666ba1e3498
trentmkelly/LessWrong-43k
LessWrong
A Layman’s Guide to Recreational Mathematics Videos This is a crosspost from my personal website.  I have watched a lot of maths videos on the internet. The medium of YouTube is quite well suited to maths; maths books are frequently either boring or are really about maths history/psychology/sociology. People sometime...
3c1dcd94-ba57-4d43-a565-0a28bc23bb6e
trentmkelly/LessWrong-43k
LessWrong
Reflections on Duo Standard Previously: Speculations on Duo Standard, Reflections on the Mythic Invitational Companion Piece: Reflections on the Mythic Invitational Frank Karsten (Channel-Fireball): The Mythic Invitational Wasn’t Perfect And It Was Still a Smashing Success There are two big questions to ask about D...
1ed81ecc-5815-4f9d-8f66-a6c7b6dc0605
trentmkelly/LessWrong-43k
LessWrong
What is Rationalist Berkey's Community Culture
e0437831-47c3-460a-af9b-6eedc0d9026f
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Direct Preference Optimization in One Minute The [Direct Preference Optimization (DPO) paper](https://arxiv.org/pdf/2305.18290.pdf) promises a more simple and efficient alternative to PPO that is able to void the reward modeling phase, and thus optimize directly for the preferences expressed in the preference data. Th...
ab9771eb-e4be-418c-b689-c6a3102a66d4
trentmkelly/LessWrong-43k
LessWrong
Question about Sociopathy/Psychopathy/ASPD I have consistently, over the course of my life, heard people describe sociopathy and related mental illnesses as being caused by a lack of empathy. This, intuitively, seems wrong, since that seems like a massively important brain function, that really ought to have a major a...
9eef79d7-ea2b-4a66-8b62-dbc1e6c0bb25
trentmkelly/LessWrong-43k
LessWrong
Costs and benefits of amniocentesis for normal pregnancies Disclaimer: No medical training. Amniocentesis is a sample of the amniotic fluid to test fetal DNA at 15-20 weeks of pregnancy. It involves inserting a long needle into the amniotic sack, which is potentially risky. But it could catch serious defects. Arguab...
6e4c3f10-138c-41a4-a25a-a42348c10214
trentmkelly/LessWrong-43k
LessWrong
LLM AGI will have memory, and memory changes alignment Summary: When stateless LLMs are given memories they will accumulate new beliefs and behaviors, and that may allow their effective alignment to evolve. (Here "memory" is learning during deployment that is persistent beyond a single session.)[1] LLM agents will ha...