id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
4ad98552-30e1-447a-bd68-0d90fae3fd14 | trentmkelly/LessWrong-43k | LessWrong | Figuring out what Alice wants, part I
This is a very preliminary two-part post sketching out the direction I'm taking my research now (second post here). I'm expecting and hoping that everything in here will get superseded quite quickly. This has obvious connections to classical machine intelligence research areas (su... |
5a06f32d-b210-48d4-8f9f-ec9d9a0523b1 | trentmkelly/LessWrong-43k | LessWrong | Crock, Crocker, Crockiest
Summary: If you declare you’re operating by Crocker’s Rules, other people are allowed to optimize their messages to you for information, not for being nice to you. The converse of Crocker’s Rules would be asking people to optimize their messages to you for being nice to you, not for informati... |
faaca772-db9c-4f27-a0f2-9e8e25d37997 | trentmkelly/LessWrong-43k | LessWrong | H-JEPA might be technically alignable in a modified form
This post is a collection of replies to Steven Byrnes' "LeCun’s “A Path Towards Autonomous Machine Intelligence” has an unsolved technical alignment problem" which wind up into a whole separate post because I suggest tweaks to the H-JEPA architecture (aka APTAMI... |
2fd41b1d-9985-49d0-91e3-0619eb0ace36 | trentmkelly/LessWrong-43k | LessWrong | An argument on animal consciousness (soliciting criticism)
Consciousness. I hope we can agree that all living people are conscious and their suffering matters.
But why stop at currently living people? We would probably also agree that earlier people, be it one year, a century, or a millennia ago, were also morally re... |
9ac29d2a-3cb1-4281-b09b-e1d867cbf936 | trentmkelly/LessWrong-43k | LessWrong | Arbital Scrape V2
I've scraped http://arbital.com as the site is unusably slow and hard to search for me.
https://drive.google.com/open?id=1b7dKhOzfMpFwngAeI8efeOzv147Lv_mx
https://mega.nz/#!SUQg1YxS!G7bmpNipcl1Ztugvr6l6sUJUu1D0gDg1-jA6oXLNEWg
https://emma-borhanian.github.io/arbital-scrape/
The scrape is locally ... |
1fa1e47e-4986-4880-bf8e-5958b0a777c5 | trentmkelly/LessWrong-43k | LessWrong | [LINK] "The current state of machine intelligence"
This article by Shivon Zilis on O'Reilly's website may be of interest. There's essentially no technical content (I think the author is a finance person with little technical expertise); it's a fairly fluffy high-level survey of the ecosystem, asking questions more lik... |
a625704d-3a6e-4d33-a0f4-aad6aa7e116c | trentmkelly/LessWrong-43k | LessWrong | Statistical checks on some social science
Simonsohn, a social scientist, investigates bad use of statistics in his field.
A few good quotes:
> The three social psychologists set up a test experiment, then played by current academic methodologies and widely permissible statistical rules. By going on what amounted to ... |
3e209fcc-cdf3-47ce-9e6f-5a7c79e69f58 | trentmkelly/LessWrong-43k | LessWrong | Why did Europe conquer the world?
For most of history, China was the center of civilization. It had the biggest cities, the most complex government, the highest quality manufacturing, the most industrial capacity, the most advanced technology, the best historical records and the largest armies. It dominated East Asia ... |
0d39b1af-6528-40ba-b5c5-b9be72acfbab | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Announcing new round of "Key Phenomena in AI Risk" Reading Group
**TLDR:**“[Key Phenomena in AI Risk](https://www.lesswrong.com/posts/mqvxR9nrXAzRr3ow9/announcing-key-phenomena-in-ai-risk-facilitated-reading)” is an 8-week-long, facilitated reading group. It is aimed at people interested in conceptual AI alignment res... |
9ecfe814-4f40-462b-b04a-cc5ad62c540a | trentmkelly/LessWrong-43k | LessWrong | Why do people commit mathematical mistakes? What are the mechanisms behind them?
I'm mostly asking this open question to those among us who are well-versed in developmental psychology (I'm mostly thinking of children) . Although, failing the actual scientific research on the topic, I guess some testable hypotheses wou... |
acad837e-12b5-4958-ac69-61df5d4216a6 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "You are not a Bayesian homunculus whose reasoning is 'corrupted' by cognitive biases.
You just are cognitive biases.
You just are attribution substitution heuristics, evolved intuitions, and unconscious learning. These make up the 'elephant' of your mind, and atop them rides a tiny 'deliberative thinking' module that ... |
0c9443cc-3355-4a3c-ac6b-866d9b54cedd | trentmkelly/LessWrong-43k | LessWrong | Panology
The study of literally everything would, logically, be called panology.
Panology Does Not (Currently) Exist
Every field in science would be a subfield of panology, if panology existed, which, as of 2024, it does not.
Since there is no "Field Of Study" officially called panology in existing academic science... |
e49244dd-7526-49c0-995c-931edb7922db | trentmkelly/LessWrong-43k | LessWrong | Expectations Influence Reality (and AI)
Edit: I want to be very clear that I'm not espousing the law of attraction. Instead, I'm saying that expectations influence reality—but always indirectly, through our actions and behavior (as informed by our beliefs). The most accurate level of description will always follow the... |
bcb69ce7-7c23-4890-b434-b4d867128367 | trentmkelly/LessWrong-43k | LessWrong | Attempts to Debias Hindsight Backfire!
(Content note: A common suggestion for debiasing hindsight: try to think of many alternative historical outcomes. But thinking of too many examples can actually make hindsight bias worse.)
Followup to: Availability Heuristic Considered Ambiguous
Related to: Hindsight Bias
I.
... |
3bd7a38e-19ac-493e-a0bf-1889e83154ac | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | A course for the general public on AI
Hi everyone!
I'm came across this Coursera course on AI for the general public. It's called [AI For Everyone](https://www.coursera.org/learn/ai-for-everyone), and it's taught by Andrew Ng: cofounder of Coursera, and professor of Computer Science at Stanford.
I think it could be ... |
b0b3a78e-b799-4244-ab59-7bff85a30d72 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Latent Skill Embedding for Personalized Lesson Sequence Recommendation
1 Introduction
---------------
The popularity of online education platforms has soared in recent years. Companies like Coursera and EdX offer Massive Open Online Courses (MOOCs) that attract millions of students and high-calibre instructors. Kha... |
46fbb71e-3f5b-4e44-9f50-647169cd4639 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Using Self-Supervised Learning Can Improve Model Robustness and Uncertainty.
1 Introduction
---------------
Self-supervised learning holds great promise for improving representations when labeled data are scarce. In semi-supervised learning, recent self-supervision methods are state-of-the-art (rotnet; exemplar\_ne... |
c9d56c25-68d9-4ad9-9eed-603f7a2a98fc | StampyAI/alignment-research-dataset/blogs | Blogs | Ariel Procaccia on economics and computation
 Ariel Procaccia is an assistant professor in the [Computer Science Department](http://www.cs.cmu.edu/) at [Carnegie Mellon University](http://www.cmu.edu/). He received his Ph... |
eead18c3-093c-4a3d-92b9-8f576f35f5ac | trentmkelly/LessWrong-43k | LessWrong | A puzzle concerning CS major vs. engineering major salaries
In the process of investigating the relative merits of majoring in computer science versus various engineering specialties, I came across the following puzzle, which I've been unable to solve. Maybe one of you can.
The puzzle is that:
* Computer science ma... |
a04c0a31-581b-411d-90c3-78656de0bee5 | trentmkelly/LessWrong-43k | LessWrong | Gossip
Epistemic status: an attempt to model gossip, not based on research. confidant that this describes real dynamics, though there's room for refinement.
Gossip requires at least 3 parties - the Gossiper, the Target (who's being gossiped about), and the Recipient (who's being gossiped to).
The two basic questions... |
40bf8c07-a8b4-428d-8d2b-c03e0ca063fb | trentmkelly/LessWrong-43k | LessWrong | Notes on Henrich's "The WEIRDest People in the World" (2020)
Cross-posted to the Effective Altruism Forum (but with a brand new and improved structure)
I recently finished reading Henrich's 2020 book The WEIRDest People in the World: How the West Became Psychologically Peculiar and Particularly Prosperous. I would hi... |
76aba058-a7d6-49eb-8412-1cef7c753ed7 | trentmkelly/LessWrong-43k | LessWrong | Saving Time
For the last few years, a large part of my research motivation has been directed at trying to save the concept of time—save it, for example, from all the weird causal loops created by decision theory problems. This post will hopefully explain why I care so much about time, and what I think needs to be fixe... |
f43e632a-08bb-426d-974a-6f18ef8f3fa9 | trentmkelly/LessWrong-43k | LessWrong | SAE Training Dataset Influence in Feature Matching and a Hypothesis on Position Features
Abstract
Sparse Autoencoders (SAEs) linearly extract interpretable features from a large language model's intermediate representations. However, the basic dynamics of SAEs, such as the activation values of SAE features and the enc... |
a6dacf06-4c48-497e-8107-96fb324c9c56 | trentmkelly/LessWrong-43k | LessWrong | Chapter 104: The Truth, Pt 1, Riddles and Answers
June 13th, 1992.
It was the last week of school in Hogwarts, and Professor Quirrell was still alive, barely. The Defense Professor himself would be in a healer's bed, this day, as he'd been for almost the last week.
Hogwarts tradition said that exams were given in th... |
cfeed49d-03b5-4db5-91f5-43032f6eab2b | trentmkelly/LessWrong-43k | LessWrong | Supermen of the (Not so Far) Future
Despite being fairly well established as a discipline, genetics is a science that has yet to reach its potential, both policymakers and the general population are extremely skeptical of it and, as a consequence, our society has setup several barriers to prevent its flourishing.
In ... |
33467738-a44a-4be5-b987-636a8fe1f574 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Deep Reinforcement Learning with Feedback-based Exploration
I Introduction
---------------
Contemporary control engineering is adopting the data-driven domain where high-dimensional problems of increasing complexity are solved, even if these are intractable from a classic control perspective.
Learning algorithms, i... |
c3c47c3a-f155-4f71-b944-4e311680e185 | trentmkelly/LessWrong-43k | LessWrong | Outline of possible Singularity scenarios (that are not completely disastrous)
Suppose we could look into the future of our Everett branch and pick out those sub-branches in which humanity and/or human/moral values have survived past the Singularity in some form. What would we see if we then went backwards in time and... |
18a9fe2b-7589-48eb-9573-6ddd3081cbaa | trentmkelly/LessWrong-43k | LessWrong | Negative Polyamory outcomes?
Related article: Polyhacking
Although polyamory isn't one of the "official" topics of LW interest (human cognition, AI, probability, etc...), this is the only community I'm part of where I expect a sufficiently high number of members to have experience with it to give useful feedback. ... |
0b915edf-1eae-41a5-ab53-f2ae21f32421 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | A compressed take on recent disagreements
An underlying generator of many recent disagreements appears to be differing locations of the disagreeing individuals on the [Yudkowsky-Hanson spectrum](http://intelligence.org/files/AIFoomDebate.pdf). That is, differences in individuals' distributions over the background vari... |
dea450ea-c717-41e5-afca-7264abc1b752 | trentmkelly/LessWrong-43k | LessWrong | Education on My Homeworld
I appreciate Eliezer Yudkowsky going public with where he came from. It has given me the courage to speak about the planet I'm from.
Sometimes people ask me how I know so much about such a wide range of topics. It flatters my ego to say the answer is self-study, but while that is technically... |
b6174321-acce-4738-9b07-89b68cf81b68 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "This is an extension of a comment I made that I can't find and also a request for examples. It seems plausible that, when giving advice, many people optimize for deepness or punchiness of the advice rather than for actual practical value. There may be good reasons to do this - e.g. advice that sounds deep or punchy mi... |
93daf313-498b-4613-b895-d05e164e57a7 | StampyAI/alignment-research-dataset/arbital | Arbital | Proof of Löb's theorem
The intuition behind the proof of Löb's theorem invokes the formalization of the following argument:
Let $S$ stand for the sentence "If $S$ is provable, then I am Santa Claus".
As it is standard, when trying to prove a If...then theorem we can start by supposing the antecedent and checking tha... |
525f931e-440b-4cfd-8776-76a09fc62658 | trentmkelly/LessWrong-43k | LessWrong | The Blue-Minimizing Robot
Imagine a robot with a turret-mounted camera and laser. Each moment, it is programmed to move forward a certain distance and perform a sweep with its camera. As it sweeps, the robot continuously analyzes the average RGB value of the pixels in the camera image; if the blue component passes a c... |
fae8bf1e-4f11-4355-8f83-7dbf898aab3e | trentmkelly/LessWrong-43k | LessWrong | [SEQ RERUN] Torture vs. Dust Specks
Today's post, Torture vs. Dust Specks was originally published on 30 October 2007. A summary (taken from the LW wiki):
> If you had to choose between torturing one person horribly for 50 years, or putting a single dust speck into the eyes of 3^^^3 people, what would you do?
Di... |
ae467651-b25a-4eed-831e-8f34700c627c | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "Last Sunday I and a small group of volunteers ran an experiment. We got a bunch of CFAR alumni together and asked half of them to tutor the other half in math (in a broad sense, so including computer science) while paying close attention to what was happening in the minds of the tutees, focusing on issues like where t... |
5935707a-b7f5-46b5-b70a-29e0b1e341c3 | trentmkelly/LessWrong-43k | LessWrong | To what degree do you model people as agents?
The idea for this post came out of a conversation during one of the Less Wrong Ottawa events. A joke about being solipsist turned into a genuine question–if you wanted to assume that people were figments of your imagination, how much of a problem would this be? (Being told... |
5656caeb-46e7-4d4b-97df-4ca3d0e7845e | trentmkelly/LessWrong-43k | LessWrong | Play for a Cause
Some of you have been trying to raise money for the Singularity Institute, and I have an idea that may help.
The idea is to hold public competitions on LessWrong with money going to charity. Agree to a game and an amount of money, then have each player designate a charity. After the game, each player... |
924475ea-86a1-4121-918f-4777009aa479 | trentmkelly/LessWrong-43k | LessWrong | Week One of Studying Transformers Architecture
I asked Claude already, but he's kind of a suckup, so if any experts find themselves reading, please tell me if anything here is super wrong.
I’ve been writing a lot about the meta level, which I think can be a trap for blogs in general, and blogs in the rationalist dias... |
5e856499-02e1-4c15-bbcb-17e7704c964e | trentmkelly/LessWrong-43k | LessWrong | Chapter 26: Noticing Confusion
Yakka foob mog. Grug pubbawup zink wattoom gazork. Chumble spuzz J. K. Rowling.
----------------------------------------
Professor Quirrell's office hours consisted of 11:40 to 11:55 AM on Thursday. That was for all of his students in all years. It cost a Quirrell point just to knock o... |
4e4e8923-90a0-4f8e-9084-bc0dccb0579e | trentmkelly/LessWrong-43k | LessWrong | Six (and a half) intuitions for SVD
The long-awaited[1] sequel to my "Six (and a half) intuitions for KL divergence" is finally here!
Thanks to the following people for feedback: Denizhan Akar, Rudolf Laine, Simon Lermen, Aryan Bhatt, Spencer Becker-Kahn, Arthur Conmy, and anonymous members of my ARENA cohort (-:
I... |
acd37665-6e22-488e-8578-443990beb134 | trentmkelly/LessWrong-43k | LessWrong | LWers on last.fm
Since several people spent a while comparing musical taste on IRC, I thought it was worth creating a group for those of us on last.fm:
Less Wrong |
a5ff25a7-6ba6-451b-b649-aaf6952213ec | StampyAI/alignment-research-dataset/arbital | Arbital | Problem of fully updated deference
The problem of 'fully updated deference' is an obstacle to using [moral uncertainty](https://arbital.com/p/7s2) to create [corrigibility](https://arbital.com/p/45).
One possible scheme in [AI alignment](https://arbital.com/p/2v) is to give the AI a state of [https://arbital.com/p/-... |
67fcf9d0-a225-43dc-b78f-843b48a5a42d | trentmkelly/LessWrong-43k | LessWrong | Meetup : Baltimore Weekly Meetup
Discussion article for the meetup : Baltimore Weekly Meetup
WHEN: 24 July 2016 08:00:00PM (-0400)
WHERE: 1726 reisterstown road pikesville maryland 21208
Pikesville DoubleTree - Hilton, at the restaurant / bar or at one of the tables outside.
Discussion article for the meetup : Ba... |
c5d27f49-985b-4392-a9ed-0255b0bf9744 | trentmkelly/LessWrong-43k | LessWrong | A Qualitative Case for LTFF: Filling Critical Ecosystem Gaps
The longtermist funding ecosystem needs certain functions to exist at a reasonable scale. I argue LTFF should continue to be funded because we're currently one of the only organizations comprehensively serving these functions. Specifically, we:
* Fund earl... |
9d1d7c93-6f23-42f0-a9ee-0d1fb671ca49 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | The OpenAI playground for GPT-3 is a terrible interface. Is there any great local (or web) app for exploring/learning with language models?
I'm thinking something that you add your API keys, and it has a pleasant interface for writing prompts, seeing results, and storing all the prompts and outputs for your reference.... |
cdfc0a11-c976-4ff8-bafd-0a1448d4ad38 | trentmkelly/LessWrong-43k | LessWrong | How to Absorb a Shared Success Script (while also thinking you're living without one)
(Cross-posted from my roam-blog)
Productivity Origin Story
Growing up, the one thing I wanted more than anything else was super powers. It wasn't until the middle of 6th grade that I finally gave up on trying to nurture my not quit... |
a5e0b96c-65ba-4999-9b24-855524edbac1 | trentmkelly/LessWrong-43k | LessWrong | AXRP Episode 7.5 - Forecasting Transformative AI from Biological Anchors with Ajeya Cotra
Audio unavailable for this episode.
This podcast is called AXRP, pronounced axe-urp and short for the AI X-risk Research Podcast. Here, I (Daniel Filan) have conversations with researchers about their papers. We discuss the pape... |
f6b7b596-0240-47ee-9068-142668b0ba71 | trentmkelly/LessWrong-43k | LessWrong | What's Wrong With the Simulation Argument?
In LessWrong contributor Scott Alexander's essay, Espistemic Learned Helplessness, he wrote,
> Even the smartest people I know have a commendable tendency not to take certain ideas seriously. Bostrom’s simulation argument, the anthropic doomsday argument, Pascal’s Mugging – ... |
5cabefb4-6735-43e6-98e2-09c12f3a33c2 | trentmkelly/LessWrong-43k | LessWrong | Postel’s Principle as Moral Aphorism
[All the usual disclaimers. Wanders dangerously close to moral relativism. Cross-posted from Grand, Unified, Empty.]
I.
Postel’s Principle (also known as the Robustness Principle) is an obscure little guideline somewhat popular among computer programmers, in particularly those wo... |
06009684-9d10-43f0-823a-313beff6fc32 | trentmkelly/LessWrong-43k | LessWrong | Discord space for people with FTX clawbacks/claims request
tl:dr; there is a support group for those dealing with FTX legal proceedings on https://discord.gg/X67NztVg
A while back, I asked this question looking for anyone navigating the FTX "clawbacks" and related matters. A few folks got in touch, but it's become cl... |
0db0e9d8-e7a0-4fbf-89d2-f7688154e377 | trentmkelly/LessWrong-43k | LessWrong | The Promise and Peril of Finite Sets
Has this ever happened to you?
> But how can we decide whether this {species/moral theory/consensus node/quantum world} is actually 2 different things, or just 1 thing? Or does it even count at all?
The underlying cause of such dilemmas is that somebody, somewhere in the setup of... |
955643a0-9fcb-41b6-825b-3cf29fe1a088 | trentmkelly/LessWrong-43k | LessWrong | If no near-term alignment strategy, research should aim for the long-term
This is a small point about alignment strategy that gets mentioned occasionally but hasn't been stated explicitly as far as I can tell [1].
If there are no paths to alignment that can be implemented in the near-term, research should focus on b... |
62d55503-e53e-45b7-9d22-d18c3477ac38 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | A broad basin of attraction around human values?
Followup to: [Morality is Scary](https://www.lesswrong.com/posts/y5jAuKqkShdjMNZab/morality-is-scary), [AI design as opportunity and obligation to address human safety problems](https://www.lesswrong.com/posts/vbtvgNXkufFRSrx4j/three-ai-safety-related-ideas)
In [Corri... |
961ac046-09cb-4701-8690-4e59b2a37776 | trentmkelly/LessWrong-43k | LessWrong | [Link] Truth-telling is aggression in zero-sum frames (Jessica Taylor)
https://unstableontology.com/2019/09/10/truth-telling-is-aggression-in-zero-sum-frames/
Excerpt:
> If we adopt a frame that says that unusual social plots are actions that are against someone (which is a zero-sum frame), this leads to the conclu... |
0d71ad5d-6d44-4f01-9e30-8881960123a1 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | AI-Written Critiques Help Humans Notice Flaws
This is a linkpost for a recent paper from the OpenAI alignment team (disclaimer: I used to work with this team). They summarize their results as:
> We trained “critique-writing” models to describe flaws in summaries. Human evaluators find flaws in summaries much more of... |
48deafbd-6a6c-4f5e-a0ba-a27f8e356087 | trentmkelly/LessWrong-43k | LessWrong | What makes us think _any_ of our terminal values aren't based on a misunderstanding of reality?
Let's say Bob's terminal value is to travel back in time and ride a dinosaur.
It is instrumentally rational for Bob to study physics so he can learn how to build a time machine. As he learns more physics, Bob realizes that... |
f593d231-18cc-48c1-a87e-a3448e136865 | trentmkelly/LessWrong-43k | LessWrong | Purple Lipstick
Wilfred's date had stood him up four times out of four. Wilfred opened his notebook even though he never actually got any work done. He didn't know why he kept coming back to Nostalgia Cafe.
The entry bell jingled. Susan's suit was rumpled. She seemed tense, like a tiger with poachers encroaching on i... |
bce7b26f-72e0-4371-bb47-0625fe48ba1e | trentmkelly/LessWrong-43k | LessWrong | One argument in favor of limited life spans
It's the only absolutely reliable way of getting rid of bad leaders.
This might not be a good enough reason to oppose longevity tech, but I don't think it's easily disposed of. |
de876b34-a12e-408f-8544-49bca20427bf | trentmkelly/LessWrong-43k | LessWrong | Public Positions and Private Guts
In this post I lay out a model of beliefs and communication that identify two types of things we might think of as ‘beliefs,’ how they are communicated between people, how they are communicated within people, and what this might imply about intellectual progress in some important fiel... |
b3c17718-d724-4915-9d77-f597c9fcf00e | trentmkelly/LessWrong-43k | LessWrong | Meaning Machines
Introduction
This post is an attempt to build up a very sparse ontology of consciousness (the state space of consciousness). The main goal is to suggest that a feature commonly considered to be constitutive of conscious experience--that of intentionality or aboutness--is actually a kind of emergent il... |
267e99fc-f120-4758-b62a-5919775033d3 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Urbana-Champaign: Reconstituting
Discussion article for the meetup : Urbana-Champaign: Reconstituting
WHEN: 31 August 2014 02:00:00PM (-0500)
WHERE: 206 S. Cedar St, 61801
In the puddles and pools formed by the fall rains, meetup groups that seemed dead end their arid hibernation and emerge from the earth.... |
37d8fdbf-1188-43b7-8e6c-cb5ff9d1107a | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Philosophical Cyborg (Part 2)...or, The Good Successor
*This post is part of the output from AI Safety Camp 2023’s Cyborgism track, run by Nicholas Kees Dupuis - thank you to Nick, AISC organizers & funders for their support.*
TL;DR
=====
This post follows up on the [cyborgism](https://www.lesswrong.com/posts/bxt7... |
b3688124-8921-4432-9e9c-0732389250a8 | trentmkelly/LessWrong-43k | LessWrong | Should we exclude alignment research from LLM training datasets?
This is a companion post to Keeping content out of LLM training datasets, which discusses the various techniques we could use and their tradeoffs. My intention is primarily to start a discussion, I am not myself very opinionated on this.
As AIs become m... |
eca25451-1e35-4a13-aefe-bec413dca6a0 | trentmkelly/LessWrong-43k | LessWrong | Self location for LLMs by LLMs: Self-Assessment Checklist.
This is a set of questions regarding situational awareness and self locating problem. These questions are from the perspective of the AI. Purpose of these questions is to guide and/or pose targets for an AI to gain understanding of its situation, for the AIs w... |
56052b36-622a-4e61-8fe8-bbb4b3330888 | trentmkelly/LessWrong-43k | LessWrong | [Link] How I Escaped The Darkness of Mental Illness
A deeply personal account by aspiring rationalist Agnes Vishnevkin, who shares the broad overview of how she used rationality-informed strategies to recover from mental illness. She will also appear on the Unbelievers Radio podcast today live at 10:30 PM EST (-5 UTC)... |
3933bed7-f3de-4be6-b4ef-bea79a820e9d | trentmkelly/LessWrong-43k | LessWrong | Be Wary of Thinking Like a FAI
I recently realized that, encouraged by LessWrong, I had been using a heuristic in my philosophical reasoning that I now think is suspect. I'm not accusing anybody else of falling into the same trap; I'm just recounting my own situation for the benefit of all.
I actually am not 100% sur... |
b9324fff-4b28-4168-afe3-d5c205a3b37c | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Counterfactually uninfluenceable agents
.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... |
b397ca82-b532-44de-9ef6-c61caf393316 | trentmkelly/LessWrong-43k | LessWrong | Computability and Complexity
Perhaps the most difficult problem in computer science is to gain some understanding of what a program does without having to run it. This might seem surprising to you. Why can't we simply look at the source code or some other description of the program and "see" what it does?
If so, then... |
18548463-8cb8-48ae-91b8-5ea6a9699853 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Urbana-Champaign: Games
Discussion article for the meetup : Urbana-Champaign: Games
WHEN: 02 March 2014 12:00:00PM (-0800)
WHERE: 40.110430, -88.223784
Available games will include Wits and Wagers, Zendo, Cards Against Humanity, Pandemic, probably Flux, and the closed beta of Dystheism, a cooperative multi... |
44425805-2ff3-4a8c-8354-0b887819289e | trentmkelly/LessWrong-43k | LessWrong | Getting Feedback by Restricting Content
Sivers just posted an important point about getting feedback, to get feedback on a post, present only one idea at a time.
Original post here http://sivers.org/1idea ; Hacker News comments http://news.ycombinator.com/item?id=964183 ; my post on it http://williambswift.blogspot.c... |
7b42f0df-47c3-4ff9-93a0-726e534c72b5 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Garden grove meetup
Discussion article for the meetup : Garden grove meetup
WHEN: 22 May 2012 07:18:15PM (-0700)
WHERE: Brookhurst and Garden Grove 10130 Garden Grove Blvd, Garden Grove, CA 92844
At Genki Living. For Snacks and Drinks with light discussion on how to properly have a meetup and what to meetu... |
36d027c4-3ff5-47d3-b5a3-fdb25259e0ba | trentmkelly/LessWrong-43k | LessWrong | This Year I Tried To Teach Myself Math. How Did It Go?
Cross-posted from my substack
Summary:
This goes over a year of self teaching, the topics of math I was able to learn, how I view the best way to learn math and what my general view of math as a whole is with my limited current knowledge on the subject. At the e... |
9187d0c1-6373-48a4-82a0-94992602181e | trentmkelly/LessWrong-43k | LessWrong | Is there a practitioner's guide for rationality?
Hi everyone, I'm new to the community, and am currently working my way through the sequences — yes, all of them.
In the introduction to the first book of Rationality A-Z, Eliezer says:
> I didn’t realize that the big problem in learning this valuable way of thinking w... |
0e79c678-ffe0-4bbf-a8d8-ae6d5245ce32 | trentmkelly/LessWrong-43k | LessWrong | Sea Change
My sensors detected the villagers' approach before any of Prospera 2's human residents noticed them—a cluster of heat signatures moving slowly across the causeway, their vital signs indicating elevated stress levels and sustained malnutrition. Through the compound's surveillance system, I watched Dr. Sarah ... |
b2dc9de3-b214-4242-9a88-b29105e43125 | trentmkelly/LessWrong-43k | LessWrong | Upcoming CFAR events: Lower-cost bay area intro workshop; EU workshops; and others
For anyone who's interested:
CFAR is trying out an experimental, lower-cost, 1.5-day introductory workshop Oct 25-26 in the bay area. It is meant to provide an easier point of entry into our rationality training. If you've been think... |
ebc4fdf7-c7a1-47f3-8edd-30fd45f84c5a | trentmkelly/LessWrong-43k | LessWrong | Mathematics As Physics
Introduction
Aside from the lowest levels of physics itself, effectively everything that humanity has studied scientifically has been found to reduce to smaller physical parts. This idea is strongly backed by many, many, observations. Because this idea is strongly backed by evidence (which I wil... |
df7b6de1-2789-407d-b186-1f22da88a819 | trentmkelly/LessWrong-43k | LessWrong | The 10,000-Hour Rule is a myth
This post appeared first on the EA Coaching blog.
The 10,000-Hour Rule for expertise popularized by Malcolm Gladwell is a funny case of citation telephone. Basically, the rule claims that anyone can become an expert if they spend ten thousand hours practicing. In other words, ten thousa... |
e560628e-067c-410c-a896-bd74c4aec53b | trentmkelly/LessWrong-43k | LessWrong | What new technology, for what institutions?
Thinking about designs for new technology is something of a hobby of mine. For simple examples, you can see this battery design or some of my blog posts. Even if you can't come up with something useful (and you probably can't) having a problem in mind is a good way to guide ... |
b1e9b2bb-f9ab-4fc4-b084-1a99872a863f | trentmkelly/LessWrong-43k | LessWrong | Meetup : Israel Less Wrong Meetup - Social and Board Games
Discussion article for the meetup : Israel Less Wrong Meetup - Social and Board Games
WHEN: 02 October 2014 07:00:00PM (+0300)
WHERE: Google Tel Aviv
We're going to have a meetup on Thursday, October 2nd at Google Israel's offices, Electra Tower, 98 Yigal A... |
d9aa9bec-3edc-47a7-ad51-0325d04d6c59 | trentmkelly/LessWrong-43k | LessWrong | Reflections on my first year of AI safety research
Last year, I wrote a post about my upskilling in AI alignment. To this day, I still get people occasionally reaching out to me because of this article, to ask questions about getting into the field themselves. I’ve also had several occasions to link people to the arti... |
8a3c21ed-e349-4fda-a8ce-919428956428 | trentmkelly/LessWrong-43k | LessWrong | Lies and Secrets
My intuition says presenting bad facts or pieces of reasoning is wrong, but withholding good facts or pieces of reasoning is less wrong. I assume most of you agree.
This is a puzzle, because on the face of it, the effect is the same.
Suppose the Walrus and the Carpenter are talking of whether pigs h... |
0950539b-1138-42b1-a6a7-c425f4da0856 | StampyAI/alignment-research-dataset/blogs | Blogs | New papers on reflective oracles and agents
We recently released two new papers on reflective oracles and agents.
The first is “[Reflective oracles: A foundation for classical game theory](https://intelligence.org/files/ReflectiveOracles.pdf),” by Benja Fallenstein, Jessica Taylor, and Paul Christiano.
[![reflecti... |
b8e9ec69-e933-4e68-8735-cf042067d7d8 | trentmkelly/LessWrong-43k | LessWrong | Applying reinforcement learning theory to reduce felt temporal distance
(cross-posted from my blog)
It is a basic principle of reinforcement learning to distinguish between reward and value, where the reward of a state is the immediate, intrinsic desirability of the state, whereas the value of the state is proportion... |
53d93639-28bc-4b71-9641-19c75aa45f72 | trentmkelly/LessWrong-43k | LessWrong | What is narrow value learning?
Ambitious value learning aims to achieve superhuman performance by figuring out the underlying latent "values" that humans have, and evaluating new situations according to these values. In other words, it is trying to infer the criteria by which we judge situations to be good. This is pa... |
17582ff7-6b6c-442e-8bbb-2f41fa5d1a95 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Tucson Meetup
Discussion article for the meetup : Tucson Meetup
WHEN: 12 October 2013 02:00:00PM (-0700)
WHERE: 943 E University Blvd #191, Tucson, AZ
This is a first meetup for Tucson. We will meet at a coffee shop near the U of A, "Caffe Luce". As this is a first meeting we will be doing some introduct... |
ce14f26b-9892-48fe-b301-b792372b8469 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post3786
A putative new idea for AI control; index here . I've previously talked about stratified indifference/learning . In this short post, I'll try and present the idea, as simply and clearly as possible. The truth is yet written Assume the AI maximising a utility W , but is unsure whether W is equal to utility... |
0416cf62-287e-476e-8acd-66d384cba40c | trentmkelly/LessWrong-43k | LessWrong | What Money Cannot Buy
Paul Graham:
> The problem is, if you're not a hacker, you can't tell who the good hackers are. A similar problem explains why American cars are so ugly. I call it the design paradox. You might think that you could make your products beautiful just by hiring a great designer to design them. But ... |
81f6257b-2e59-44a0-9470-96c55229a6c4 | trentmkelly/LessWrong-43k | LessWrong | Seeking rationality-related blogs
I am working on a rationality blog aggregator.
You can help by responding with some rationality-related blogs that you personally find valuable.
I'll announce here when I'm ready to release it publicly. |
692ff935-2003-4061-aa2b-d8d8c7a8ffad | trentmkelly/LessWrong-43k | LessWrong | There are no coherence theorems
[Written by EJT as part of the CAIS Philosophy Fellowship. Thanks to Dan for help posting to the Alignment Forum]
Introduction
For about fifteen years, the AI safety community has been discussing coherence arguments. In papers and posts on the subject, it’s often written that there ex... |
1d4808e2-5942-4eda-be52-7455f855278a | trentmkelly/LessWrong-43k | LessWrong | Remote Login For Turnkey Devices?
I run my rhythm stage setup on a Raspberry Pi. Mostly this works well: I plug it in, my code and init scripts bring everything up in the right order, in about a minute it's taking in audio and MIDI and putting out audio. If it doesn't work I turn it off and on again and then it does.
... |
3cb7d827-5d18-4971-ac2e-f3c26f92bdeb | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | UK policy and politics careers
*Disclaimer: this is not a comprehensive guide, just notes of the answers I gave in recent conversations I have had. It was suggested it could be useful to share. This is very UK focused.*
*My notes are available in [this Google Doc](https://docs.google.com/document/d/1769kGV1Sisw4u0cz-... |
8662d3ad-d75e-4741-8a4d-7d6c9fef32a7 | trentmkelly/LessWrong-43k | LessWrong | "What if we could redesign society from scratch? The promise of charter cities." [Rational Animations video]
In this video, we describe a speculative but promising idea to boost economic growth: giving cities the ability to write some of their own laws and design their own institutions, independently from their paren... |
252d5c0d-3695-4460-970b-2b04ab6a2b52 | trentmkelly/LessWrong-43k | LessWrong | COVID Strategy 101
Introduction:
Over the past six months of the COVID epidemic a sharp divide has emerged.
In one camp, most public health experts believe that the lockdowns in place are keeping us from unmitigated disaster and millions of deaths. In the other camp, a disparate group of scientists have argued that t... |
ca030026-4bb7-493f-8db4-7dee5b381a58 | trentmkelly/LessWrong-43k | LessWrong | Debate experiments at The Curve, LessOnline and Manifest
I like debate. I have done for years. So I have been slowly trying to improve it. Here is a set of theories I had and things, experiments I've run so far.
Theory: Any debates are good.
Are any debates actually good at all? Should I give up?
Test: Watch differ... |
817aafc5-f934-4f8f-8fa6-6e4014a49837 | trentmkelly/LessWrong-43k | LessWrong | My Dating Plan ala Geoffrey Miller
How should a rationalist approach dating? - I have just finished Geoffrey Miller's excellent book Mate, a reductionist, signalling-heavy guide to how to attracting women for men. This post lays my "dating plan" a la Miller. I post for constructive comments and to give an example of a... |
51c192f1-ea67-406e-885a-ca377a8c6eb6 | trentmkelly/LessWrong-43k | LessWrong | Open Thread, Aug. 8 - Aug 14. 2016
If it's worth saying, but not worth its own post, then it goes here.
----------------------------------------
Notes for future OT posters:
1. Please add the 'open_thread' tag.
2. Check if there is an active Open Thread before posting a new one. (Immediately before; refresh the li... |
b0413ee3-4ba6-4b3d-a7d0-6d103d17752d | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | Of data scientists and AI from critique to reflective practice (Mario Sosa Hidalgo)
uh
and if you have any questions
during the presentation please use the
raise
your hand button that you can find the
controls on msteams
if you don't find it you can type your
question also in chat
and in the second half hour we will ... |
beeda1d2-d390-4d00-a3ec-12e16bc8ba6d | trentmkelly/LessWrong-43k | LessWrong | Diaspora roundup thread, 23rd June 2016
Guidelines: Top-level comments here should be links to things written by members of the rationalist community, preferably that have some particular interest to this community. Self-promotion is totally fine. Including a brief summary or excerpt is great, but not required. Genera... |
8d3ab47b-8310-47ca-b75c-309cc8977158 | trentmkelly/LessWrong-43k | LessWrong | Hypotheses For Dualism
In this post I present the first few hypotheses that I can think for why people insist on a metaphysical aspect to consciousness, and develop one in some detail: a "reality is simulated" hypothesis.
Please contribute your own hypotheses. Why is there a persistent belief that consciousness is m... |
d9df3409-4123-46f8-9db8-6b7a4c351653 | trentmkelly/LessWrong-43k | LessWrong | The Emotional Type of a Decision
Our minds have built-in tools that we use intuitively to evaluate decisions as we go about our day-to-day lives. Sometimes, making a decision can feel particularly challenging on an emotional level. This can lead to procrastination in making the decision or just going with the default ... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.