id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
c1d70543-3885-4d5b-bc18-e01412cb00e5 | trentmkelly/LessWrong-43k | LessWrong | Reward large contributions?
Sometimes I want people to get $$$ for their contributions.
When someone posts a long, exceptional sequence? I don't get that feeling.
When folks working on the site itself roll out a sweet new feature? I don't get that feeling.
But when https://www.lesswrong.com/users/lsusr creates a fu... |
3d5bceb4-6622-4ff8-b577-c3d21a0d301d | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Half-baked alignment idea
I'm trying think through various approaches to AI alignment, and so far this is the one I came up with that I like best. I have not read much of the literature, so please do point me if this has been discussed before.
What if we train an AI agent (ie, reinforcement learning) to survive/thr... |
f0c28fa0-8c62-4b14-8fb6-8754660e86f0 | trentmkelly/LessWrong-43k | LessWrong | Slowing AI: Interventions
Disclaimer: this post is underdeveloped and doesn't have the answers. Hopefully a future version will be valuable, but I'm mostly posting this to help facilitate brainstorms and dialogues in my personal conversations.
Affordances
What important actions related to slowing AI could actors tak... |
715a8e1e-e7d5-4b0a-af9c-499b8dd71763 | trentmkelly/LessWrong-43k | LessWrong | Google Search as a Washed Up Service Dog: "I HALP!"
As is customary, I whined (on IRC) about Google Search being unhelpful or actively counterproductive with the results these days and how it is not what it used to be. Specifically, I wanted to know something very niche, why many LTE modems sometimes take too long to ... |
fa159e99-5bd5-48f3-bf32-7784be2b71c9 | trentmkelly/LessWrong-43k | LessWrong | How important is it that LW has an unlimited supply of karma?
Question
LessWrong users can up/downvote posts and comments, which then receive a karma boost (capped by the voters own karma). There is no limit to how many different posts and comments one can do this to. In this sense there is an unlimited supply of karm... |
a03eb0ca-79c6-4c1b-80fc-0b87b7f30e11 | trentmkelly/LessWrong-43k | LessWrong | What do you do to deliberately practice?
Also related: Humans Are Not Automatically Strategic
At the beginning of Tyler Cowen and David Gross's recent book, Talent, they mention an interview question Tyler used for new hires:
> Daniel recalls that he first learned from Tyler this question for prospective hires: "Wha... |
69031e92-eb94-4eae-8379-608966d59247 | trentmkelly/LessWrong-43k | LessWrong | An Akrasia Anecdote
About a month ago I committed myself to an anti-akrasia resolution, inspired by Yvain and ZM. I won't repost the resolution here, but if you want to see it, click the first link. The essence of my resolution was to commit myself to practice math to prepare for graduate school in the fall. The re... |
78936414-28e2-490a-b25b-45aa498b6855 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | How to find AI alignment researchers to collaborate with?
I am an AI researcher. I have a personal interest in AI alignment, because obviously. However I am not getting paid to do research on AI alignment, so I have to do it as a hobby.
Writing papers and submitting them to conferences is a full-time job and highly ... |
6474226d-59a0-49e6-bfc3-ae00afc7c40b | StampyAI/alignment-research-dataset/arxiv | Arxiv | The MineRL 2019 Competition on Sample Efficient Reinforcement Learning using Human Priors
1 Competition description
--------------------------
###
1.1 Background and impact
Many of the recent, most celebrated successes of artificial intelligence (AI), such as AlphaStar, AlphaGo, OpenAI Five, and their derivativ... |
9203d5c7-8fea-407a-9bde-fd7a9dc99619 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Washington, D.C.: Fun & Games
Discussion article for the meetup : Washington, D.C.: Fun & Games
WHEN: 09 April 2017 03:30:00PM (-0400)
WHERE: Donald W. Reynolds Center for American Art and Portraiture
We will be meeting in the courtyard to hang out, play games, and engage in fun conversation.
Upcoming mee... |
6a4048d9-2001-4d65-a7a3-bb648d786368 | trentmkelly/LessWrong-43k | LessWrong | Case Study: Reading Edge's financial filings
* 1 The problem
* 2 The solution
* 2.1 Which charity?
* 2.1.1 The filings
* 2.1.2 Overview
* 2.1.3 Income
* 2.1.4 Expenses
* 3 Final thoughts
* 4 Other examples
> Abstract: an example of how to read charity filings, using Edge.org
1 The pro... |
5b2a8550-0d7f-4c05-af03-fd04ddcfaaf2 | trentmkelly/LessWrong-43k | LessWrong | Math is Subjunctively Objective
Followup to: Probability is Subjectively Objective, Can Counterfactuals Be True?
I am quite confident that the statement 2 + 3 = 5 is true; I am far less confident of what it means for a mathematical statement to be true.
In "The Simple Truth" I defined a pebble-and-bucket system for... |
480c6a4c-a303-40d6-8a18-b16d21d42901 | trentmkelly/LessWrong-43k | LessWrong | New GPT3 Impressive Capabilities - InstructGPT3 [1/2]
Summary
* InstructGPT3 (hereafter IGPT3), a better version of GPT3 has recently been released by OpenAI. This post explores its new capabilities.
* IGPT3 has new impressive capabilities and many potential uses. Among others, it can help users:
* Brainstorm
... |
5b2a119d-2c3d-4f12-86f7-7407bfffb211 | trentmkelly/LessWrong-43k | LessWrong | Welcome to Boulder SSC Meetup [Edit With Your Details]
(The following are our suggestions for what kind of information is best to include in the welcome post of your group, feel free to replace them with whatever you think is best)
What kind of events does your group usually run? What does it usually do?
How frequen... |
46419277-9ca1-4f95-865a-b5912a04e1e6 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | AI Safety Newsletter #5: Geoffrey Hinton speaks out on AI risk, the White House meets with AI labs, and Trojan attacks on language models
Welcome to the AI Safety Newsletter by the [Center for AI Safety](https://www.safe.ai/). We discuss developments in AI and AI safety. No technical background required.
Subscribe [h... |
421c9690-b76d-49bf-b8eb-d11f36f7a4df | StampyAI/alignment-research-dataset/arbital | Arbital | Metric
A **metric**, sometimes referred to as a distance function, is a [https://arbital.com/p/-3jy](https://arbital.com/p/-3jy) that defines a [real](https://arbital.com/p/-4bc) nonnegative distance between every two elements of a [set](https://arbital.com/p/3jz). It is commonly denoted by the variable $d$. In [https... |
fabc5861-1306-4324-b9dc-3ea6af551508 | StampyAI/alignment-research-dataset/blogs | Blogs | New paper: “Forecasting using incomplete models”
[](https://arxiv.org/abs/1705.04630)MIRI Research Associate Vanessa Kosoy has a paper out on issues in naturalized induction: “[Forecasting using incomplete models](https://arxiv... |
f833386d-0793-4915-b4a4-e83260e57c8a | trentmkelly/LessWrong-43k | LessWrong | LW Women: LW Online
Standard Intro
The following section will be at the top of all posts in the LW Women series.
Several months ago, I put out a call for anonymous submissions by the women on LW, with the idea that I would compile them into some kind of post. There is a LOT of material, so I am breaking them dow... |
9e9290e2-c262-45d7-a961-5e17c07a64b6 | trentmkelly/LessWrong-43k | LessWrong | Deciding on our rationality focus
I have a problem: I'm not sure what this community is about.
To illustrate, recently I've been experimenting with a number of tricks to overcome my akrasia. This morning, a succession of thoughts struck me:
1. The readers of Less Wrong have been interested in the subject of akrasia... |
fa0105ff-17f5-4661-9505-092001ec7b6d | trentmkelly/LessWrong-43k | LessWrong | Ergonomics Revisited
Continuation of: Spend Money on Ergonomics, by Kevin
Three years have elapsed since Kevin wisely told us to spend money on treating our bodies well. It may be time to check for new gadgets, to verify what has worked and what has not etc...
If you have purchased an item for this purpose, or i... |
5d02d4e8-751a-41a2-97c9-eafd2dc6fd40 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Washington, D.C.: Optical Illusions
Discussion article for the meetup : Washington, D.C.: Optical Illusions
WHEN: 02 August 2015 03:00:00PM (-0400)
WHERE: National Portrait Gallery
Crossposted from the mailing list:
We will be congregating in the courtyard between 3:00 to 3:30 p.m.; the meetup runs from 3... |
a4cb8e07-a175-4a58-9f63-13142d411843 | trentmkelly/LessWrong-43k | LessWrong | Rationalist Judo, or Using the Availability Heuristic to Win
During the sessions at the 2011 rationality minicamp, we learned that some of our biases can be used constructively, rather than just tolerated and avoided.
For example, in an excellent article discussing intuitions and the way they are formed, psychologist... |
1c3f4e6a-cb4b-4401-8134-9a3ab2d8bd2a | trentmkelly/LessWrong-43k | LessWrong | Contra Caller Signs
A contra dance stage can be a bit chaotic and noisy, especially during a dance. It's a good fit for gestures, and over time callers and musicians have settled on a few covering the main things they need to communicate. I couldn't find a listing of these anywhere, so here are the main ones:
* Clos... |
b714fb5c-cd0f-49ce-a68a-0f3140d414c8 | trentmkelly/LessWrong-43k | LessWrong | Individual AI representatives don't solve Gradual Disempowerement
Imagine each of us has an AI representative, aligned to us, personally. Is gradual disempowerment solved?[1] In my view, no; at the same time having AI representatives helps at the margin.
I have two deep reasons for skepticism.[2] Here is the first on... |
411a45f0-51f7-42f6-a84a-9634723fd700 | trentmkelly/LessWrong-43k | LessWrong | Chapter 27: Empathy
J. K. Rowling is 87% confident you will burst into flames.
Roger Bacon lived in the 13th century and is credited as one of the earliest advocates of the scientific method. Giving a scientist his experimental diary is sort of like giving a writer the pen, not of Shakespeare, but of someone who help... |
f554484d-ad63-41e9-96df-6aea8b03a056 | trentmkelly/LessWrong-43k | LessWrong | Predictive Coding has been Unified with Backpropagation
Artificial Neural Networks (ANNs) are based around the backpropagation algorithm. The backpropagation algorithm allows you to perform gradient descent on a network of neurons. When we feed training data through an ANNs, we use the backpropagation algorithm to tel... |
f76fb39f-4d22-4957-a6e5-ccbdcf9d60e0 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Non-Adversarial Imitation Learning and its Connections to Adversarial Methods
1 Introduction
---------------
Imitation learning (imitation learning (IL), Schaal, [1999](#bib.bib40); Osa et al., [2018](#bib.bib34)) and inverse reinforcement learning (inverse reinforcement learning (IRL), Ng and Russell, [2000](#bib.... |
802e18ba-4999-4ebb-a4cc-7970df53e123 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Catastrophic Risks from AI #1: Summary
*This is the first post in a sequence of posts giving an* [*overview of catastrophic AI risks*](https://arxiv.org/abs/2306.12001)*.*
Abstract
========
Rapid advancements in artificial intelligence (AI) have sparked growing concerns among experts, policymakers, and world leaders... |
e9177ce8-5869-4a78-ab85-4fd78b988d86 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Christiano decision theory excerpt
A transcribed excerpt I found interesting from the [decision theory outtake](https://80k.link/christiano-outtake) from 80,000 Hours' [second interview](https://80000hours.org/podcast/episodes/paul-christiano-a-message-for-the-future/) of Paul Christiano (starting at 14:00):
---
*... |
9f925d66-806e-43fb-a47c-bfbcf17f1303 | trentmkelly/LessWrong-43k | LessWrong | How to Better Report Sparse Autoencoder Performance
TL;DR
When presenting data from SAEs, try plotting 1/L0 against 1−Recovered Loss and fitting a Hill curve.
From Wikipedia
Long
Sparse autoencoders are hot, people are experimenting. The typical graph for SAE experimentation looks something like this. I'm using bor... |
7f5442a4-1e6e-484d-82b6-c7c841b6db75 | trentmkelly/LessWrong-43k | LessWrong | There are 125 sheep and 5 dogs in a flock. How old is the shepherd? / Math Education
|
f053a3d6-0b87-44d9-8e12-ce980fd34429 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Can independent researchers get a sponsored visa for the US or UK?
I’m an independent alignment researcher from Canada. I’m trying to figure out my long-term options (I want to move somewhere and stay there for several years).
I have some special visas as a Canadian to go to either place, but I’d need to be working ... |
fae44b3d-4792-4597-a63f-054db410c0b8 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Melbourne, practical rationality, Friday 7th October, 7pm
Discussion article for the meetup : Melbourne, practical rationality, Friday 7th October, 7pm
WHEN: Friday, 7 October 2011 07:00:00PM (+1000) (all welcome from 6PM)
WHERE: TrikeApps office, lvl 2, 55 Walsh St, West Melbourne 3003 (http://trikeapps.c... |
9cc51a6e-438c-4201-ae95-ed0a7283d8b2 | StampyAI/alignment-research-dataset/blogs | Blogs | Challenges to Christiano’s capability amplification proposal
The following is a basically unedited summary I wrote up on March 16 of my take on Paul Christiano’s AGI alignment approach (described in “[ALBA](https://ai-alignment.com/alba-an-explicit-proposal-for-aligned-ai-17a55f60bbcf)” and “[Iterated Distillation and... |
7d5c1af9-5049-4054-aa2a-18562b7a0b04 | trentmkelly/LessWrong-43k | LessWrong | The AI-box for hunter-gatherers
The following is a minor curiosity that occurred to me regarding real-world analogies to the AI-box concept.
Fundamentally, the reason that we fear a randomly-chosen super-intelligent AI is twofold:
1. It would be smarter than us, so it could outwit us no matter what its goals.
2... |
f100cac6-3867-4c89-84ad-006160e984f1 | trentmkelly/LessWrong-43k | LessWrong | Respond to what they probably meant
Edit: I didn't realize this before writing the post, but what I'm referring to is The Principle of Charity.
Story
I was confused about Node Modules, so I did a bunch of research to figure out how they work. Explaining things helps me to understand them, and I figured that others m... |
3279cb02-c20e-48ff-9f0b-42ede61fd971 | trentmkelly/LessWrong-43k | LessWrong | How to build a data center, by Construction Physics
Disclaimer: This is not work written by me. I am sharing a link I think is interesting for the LessWrong and AI Safety community.
First, what is Construction Physics? In their words:
> Construction Physics is a newsletter about the technology and economics of build... |
1d428c90-33a5-48e2-8457-101c4c3b9a3f | trentmkelly/LessWrong-43k | LessWrong | [AN #156]: The scaling hypothesis: a plan for building AGI
Alignment Newsletter is a weekly publication with recent content relevant to AI alignment around the world. Find all Alignment Newsletter resources here. In particular, you can look through this spreadsheet of all summaries that have ever been in the newslette... |
fc2d0761-bfca-4693-b262-aae142f2dc3d | trentmkelly/LessWrong-43k | LessWrong | [An email with a bunch of links I sent an experienced ML researcher interested in learning about Alignment / x-safety.]
[[FYI: I'm just copying this in and removing a few bits; apologies for formatting; I don't intend to post the attachments]]
EtA: I (embarrasingly and unfortunately) understated Richard Ngo's technic... |
b772e4bb-8d3d-4e5e-b5f2-94935993bfda | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Two New Newcomb Variants
Two Newcomb variants to add to the list of examples where optimal choice and optimal policy are diammetrically opposed. I don't think problems these exist anywhere else yet.
### 4 Boxes Problem
In a game show there are 4 transparent boxes in a row, each of which starts off with $1 inside.... |
edab469f-a7ec-4299-ad37-e9bcc4bede79 | trentmkelly/LessWrong-43k | LessWrong | Elqayam & Evans (2011) argue against a certain kind of normativism about rationality
A forthcoming edition of Behavioral and Brain Sciences will be devoted to Elqayam & Evans' (2011) critique of normativism about rationality and brief responses to it.
Abstract:
> We propose a critique of normativism, defined as the ... |
74c3fffd-b4a8-4a5c-9b25-1139ccbaf129 | trentmkelly/LessWrong-43k | LessWrong | Market norms and Social norms
Predictably Irrational
When I joined LW, I browsed through an article that recommended Predictably Irrational as a read for starters. After reading it, I decided to share my greatest use of the content.
MARKET NORMS VS SOCIAL NORMS
The idea is that there are different forms of curr... |
16671ad5-cd41-4219-8255-232043233e7c | trentmkelly/LessWrong-43k | LessWrong | On the naturalistic study of the linguistic behavior of artificial intelligence
This is draft material from the introduction of a working paper I’m about to finish: Discursive Competence in ChatGPT, Part 1: Talking with Dragons.
* * * * *
When I started playing with ChatGPT on December 1, 2022, I had no specific int... |
6bc8b32e-db4d-49e9-9ef8-2171c9dce516 | trentmkelly/LessWrong-43k | LessWrong | Novum Organum: Preface
Previously: Ruby's introduction to the Novum Organum sequence
We have used Francis Bacon's Novum Organum in the version presented at www.earlymoderntexts.com. Translated by and copyright to Jonathan Bennett. Prepared for LessWrong by Ruby.
Ruby's Reading Guide
> Novum Organum is organized as ... |
ff77ef3c-ed1a-40b8-aee7-165d2f96a2d8 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "Note: This is a personal post describing my own plans, not a post with actual research content.
Having finished my internship working with Paul Christiano and others at OpenAI, I’ll be moving to doing research at MIRI. I’ve decided to do research at MIRI because I believe MIRI will be the easiest, most convenient plac... |
3847cbd1-4a09-4e01-8b4e-8c6d9e2dfa8a | trentmkelly/LessWrong-43k | LessWrong | Turning up the Heat on Deceptively-Misaligned AI
Epistemic status: I ran the mathsy section through Claude and it said the logic was sound. This is an incoherence proof in a toy model of deceptively-misaligned AI. It is unclear whether this generalizes to realistic scenarios.
TL:DR
If you make a value function consi... |
272f60ac-1bfa-4368-bf4f-6ce60ae40c39 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Improving Generalization for Abstract Reasoning Tasks Using Disentangled Feature Representations
1 Introduction
---------------
Reasoning about abstract concepts has been a long standing challenge in machine learning. Recent work by Barret et al. [[1](#bib.bib1)] introduces a concrete problem setting for testing ge... |
b4fe8729-40ac-4189-87a3-e3e635209eed | trentmkelly/LessWrong-43k | LessWrong | I currently translate AGI-related texts to Russian. Is that useful?
I live in Saint-Petersburg, Russia, and I want to do something useful to AI-Alignment movement.
Several weeks ago I started to translate "Discussion with Eliezer Yudkowsky on AGI interventions". Now it's complete and linked to in couple thematic soci... |
574908c7-3352-4aca-bd11-7e34b60bd23d | StampyAI/alignment-research-dataset/arbital | Arbital | Pi is irrational
The number [$\pi$](https://arbital.com/p/49r) is not [rational](https://arbital.com/p/4zq).
# Proof
For any fixed real number $q$, and any [https://arbital.com/p/-45h](https://arbital.com/p/-45h) $n$, let $$A_n = \frac{q^n}{n!} \int_0^{\pi} [](https://arbital.com/p/x)^n \sin(x) dx$$
where $n!$ is th... |
c73dcab5-c183-4141-812d-f12351f1c7cb | trentmkelly/LessWrong-43k | LessWrong | Satisficers' undefined behaviour
I previously posted an example of a satisficer (an agent seeking to achieve a certain level of expected utility u) transforming itself into a maximiser (an agent wanting to maximise expected u) to better achieve its satisficing goals.
But the real problem with satisficers isn't that t... |
8fdfdf54-a379-4f93-8c89-61bdcf04078a | trentmkelly/LessWrong-43k | LessWrong | Meetup : Sydney Meetup - June
Discussion article for the meetup : Sydney Meetup - June
WHEN: 24 June 2015 06:30:00PM (+1000)
WHERE: 565 George Street, Sydney, Australia 2000
Regular location - City of Sydney RSL The restaurant on Level 2
Regular time (starting about 6:30)
Come for general socialising and interest... |
75793862-df55-4230-b979-f2c8b993408f | trentmkelly/LessWrong-43k | LessWrong | Are there AI policies that are robustly net-positive even when considering different AI scenarios?
One thing I might have noticed recently is a lot of governance on AI might require specific models of AI in society, especially over misalignment or misuse concerns.
Are there AI policies that could be robustly-net posi... |
404d52bd-6b7e-4f61-b642-5d1d60a1e318 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Book Review: Oryx and Crake
*Oryx and Crake* was written by Margarett Atwood in 2003, just a few years before the beginnings of the EA community. I first read *Oryx and Crake* shortly before I started engaging with EA ideas, and it wasn’t a coincidence. This book led me to take animal welfare and existential risk much... |
5ee33725-1782-47e9-9dc6-6b35f17f0163 | trentmkelly/LessWrong-43k | LessWrong | DC Meetup June 12th (New Location)
Sunday June 12th
Core hours: 1 PM - 3PM
Apartment #1005
3001 Veazy Terrace
Washington, DC 20008
Notes:
Core hours are 1-3, but many people are willing to stay around longer than that.
Our topic for the meeting will be "Where/why rationalists win, and where they don't". However, i... |
426fb7d1-1dc0-45c2-91a5-cbf682d25a1a | trentmkelly/LessWrong-43k | LessWrong | [Linkpost] The lethal trifecta for AI agents: private data, untrusted content, and external communication
TL;DR from the post by Simon Willison:
> The lethal trifecta of capabilities is:
>
> * Access to your private data—one of the most common purposes of tools in the first place!
> * Exposure to untrusted content... |
cb9b7eae-fc6f-49a8-a288-84c63c7e1144 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "Would kind of like to excerpt the whole post, but that feels impolite, so I'll just quote the first four paragraphs and then suggest reading the whole thing:There’s this funny thing about Goodhart’s Law, where it’s easy to say “being affected by Goodhart’s Law is bad” and “it’s better to behave in ways that aren’t as ... |
842b010a-f12d-425b-98c5-34793c1154a7 | trentmkelly/LessWrong-43k | LessWrong | The Onion Goes Inside The Biased Mind
A great piece from The Onion, inside the mind of someone arguing emotionally: Oh, No! It's Making Well-Reasoned Arguments Backed With Facts! Run!
Pure genius. |
727a0f2e-e723-4226-8425-923582116105 | trentmkelly/LessWrong-43k | LessWrong | Notes on Superwisdom & Moral RSI
These are very preliminary notes, to get the rough ideas out. There's lots of research lying around, a paper in the works, and I'm happy to answer any and all questions.
The Northstar of AI Alignment, as well as Alignment at Large, should be Superwisdom and Moral RSI (Recursive Self-I... |
7911c3af-fb0c-4e38-9881-2a1a9ef51a00 | StampyAI/alignment-research-dataset/blogs | Blogs | Why it matters if "ideas get harder to find"
*Click lower right to download or find on Apple Podcasts, Spotify, Stitcher, etc.*
There's been a fair amount of attention over the last few years on the idea that [ideas are getting harder to find](https://web.stanford.edu/~chadj/IdeaPF.pdf) (link goes to a paper by Blo... |
3d424280-aca7-478a-b52a-2dd13ff7a742 | StampyAI/alignment-research-dataset/blogs | Blogs | On the Expressivity of Markov Reward
Reward is the driving force for reinforcement learning (RL) agents. Given its central role in RL, reward is often assumed to be suitably general in its expressivity, as summarized by Sutton and Littman’s reward hypothesis:
> "...all of what we mean by goals and purposes can be we... |
bafaa733-28c2-4b87-b741-8d7755eb9597 | trentmkelly/LessWrong-43k | LessWrong | Real World Solutions to Prisoners' Dilemmas
Why should there be real world solutions to Prisoners' Dilemmas? Because such dilemmas are a real-world problem.
If I am assigned to work on a school project with a group, I can either cooperate (work hard on the project) or defect (slack off while reaping the rewards of ev... |
0796e5d8-fe1d-42b4-90e9-1a82929f6724 | StampyAI/alignment-research-dataset/special_docs | Other | Optimal policies for free recall.
THEORETICAL NOTE
Optimal Policies for Free Recall
Qiong Zhang1, 2, 3, Thomas L. Grif /uniFB01ths4, 5, and Kenneth A. Norman4, 6
1Department of Psychology, Rutgers University –New Brunswick
2Department of Computer Science, Rutgers University –New Brunswick
3Center for Cognitive Science... |
04f640c3-68b3-4141-b69d-609efcf4ad51 | trentmkelly/LessWrong-43k | LessWrong | HCH and Adversarial Questions
This is a paper I wrote as part of a PhD program in philosophy, in trying to learn more about and pivot towards alignment research. In it, I mostly aimed to build up and distill my knowledge of IDA.
Special thanks to Daniel Kokotajlo for his mentorship on this, and to Michael Brownstein... |
6c825c87-dc05-4158-9991-e4647d97f366 | trentmkelly/LessWrong-43k | LessWrong | ChatGPT banned in Italy over privacy concerns
> Italy has become the first Western country to block advanced chatbot ChatGPT.
>
> The Italian data-protection authority said there were privacy concerns relating to the model, which was created by US start-up OpenAI and is backed by Microsoft.
>
> The regulator said it... |
f1d7ff3a-e7b0-424f-87ab-c131c9163edf | trentmkelly/LessWrong-43k | LessWrong | The Drowning Child
Epistemic status: possibly very evil devil's advocacy, philospohicly unsophisticard ruminations
We are all familiar with Singer's drowning child thought experiment. I have often found it very compelling, or at least felt I ought to find it very compelling. My revealed preferences, as with those of ... |
c3c9d815-0d2b-4b2c-929e-d351b7347242 | trentmkelly/LessWrong-43k | LessWrong | Advice to junior AI governance researchers
This summer, I’m supervising some research fellows through Cambridge’s ERA AI Fellowship. The program started last week, and I’ve had conversations with about 6 fellows about their research projects & summer goals.
In this post, I’ll highlight a few pieces of advice I’ve fo... |
dcb02333-e20f-467b-a971-742e8c11c982 | trentmkelly/LessWrong-43k | LessWrong | Connectome-specific harmonic waves and meditation
TL;DR: meditation is a process for altering the harmonics of brainwaves to produce good brain states.
Pulling together a few threads:
* The theory of connectome-specific harmonic waves suggests one thing that happens in the brain is that oscillations in neuron activ... |
c2fd44b5-39ab-4d78-a946-6686303c17ac | trentmkelly/LessWrong-43k | LessWrong | Belief-conditional things - things that only exist when you believe in them
TL;DR: A little winter story about making beliefs come true. Epistemic status: this is >90% written to amuse. Not 100% though.
🎵 Do you believe... 🎵
It is the season, and I just finished watching the 2003 classic "Elf" with Will Ferrell as... |
e9b6084b-ccc7-41a8-8ea8-85f148705218 | trentmkelly/LessWrong-43k | LessWrong | Anthropic reasoning isn't magic
The user Optimization Process presented a very interesting collection of five anthropic situations, leading to seemingly contradictory conclusions where we can't conclude anything about anything.
It's an old post (which I just discovered), but it's worth addressing, because it's wrong ... |
a527a8e6-ac27-4821-9154-d10b7e27fb9c | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | 10 Reasons to Ignore AI Safety
hi Stuart Russell is an AI researcher
who I've talked about a few times on
this channel already
he's been advocating for these kinds of
safety or alignment ideas to other AI
researchers for quite a few years now
and apparently the reaction he gets is
often something like this in stage on... |
6da89fad-225a-4f61-8f2b-5b9d121ec67c | trentmkelly/LessWrong-43k | LessWrong | The First Circle
Epistemic Status: Squared
The following took place at an unconference about AGI in February 2017, under Chatham House rules, so I won’t be using any identifiers for the people involved.
Something that mattered was happening.
I had just arrived in San Francisco from New York. I was having a conversa... |
2a5610a1-1fa1-4c5c-baf5-ce774fa24f95 | trentmkelly/LessWrong-43k | LessWrong | Why do Minimal Bayes Nets often correspond to Causal Models of Reality?
Chapter 2 of Pearl's Causality book claims you can recover causal models given only the observational data, under very natural assumptions of minimality and stability[1].
In graphical models lingo, Pearl identifies a causal model of the observati... |
d4de60cb-3454-4089-9764-d623ea84ee67 | trentmkelly/LessWrong-43k | LessWrong | Is there a publicly available list of examples of frontier model capabilities?
Is there a list (something analagous to the excellent list of examples of specification gaming by Victoria Krakovna) where examples of impressive capabilities of frontier models are compiled? And if not, can you provide your own best examp... |
4d2d4886-d6f5-4525-8322-6c3084c2e1f5 | trentmkelly/LessWrong-43k | LessWrong | Exposure to Lizardman is Lethal
This is a short, experimentally-off-the-cuff post about one way in which social groups and institutions are effectively destroyed, in practice. It's fairly straightforward, but I don't try to connect all the dots (or even name all of them).
----------------------------------------
In ... |
de9b9b0f-0fad-4c8f-bf4e-9300cee97a19 | trentmkelly/LessWrong-43k | LessWrong | Accurate Models of AI Risk Are Hyperexistential Exfohazards
(Where "an exfohazard" is information which leads to bad outcomes if known by a large fraction of society.)
Let us suppose that we've solved the technical problem of AI Alignment — i. e., the problem of AI control. We have some method of reliably pointing ou... |
1fe35f6a-26ca-4b6e-bbc2-61afdb9ae5ad | trentmkelly/LessWrong-43k | LessWrong | Things roll downhill
(Response to this challenge)
I've read two things recently on similar strategies: Zvi's post on ChatGPT, and Scott Alexander's post on Redwood Research.
They both seem to have a similar strategy, to train AI to not do misaligned things by giving it a bunch of examples of misaligned things, and s... |
234c883b-9de3-4246-8859-ebbbfdda5150 | trentmkelly/LessWrong-43k | LessWrong | Notion of Preference in Ambient Control
This post considers ambient control in a more abstract setting, where controlled structures are not restricted to being programs. It then introduces a notion of preference, as an axiomatic definition of constant (actual) utility. The notion of preference subsumes possible worlds... |
b9d4ae7f-b823-42c6-914b-4806cbb65892 | trentmkelly/LessWrong-43k | LessWrong | Open Thread, October 20 - 26, 2013
If it's worth saying, but not worth its own post (even in Discussion), then it goes here.
|
be4672ec-19fc-4dc9-ab8f-d3e5ede8a52b | trentmkelly/LessWrong-43k | LessWrong | Critique of 'Many People
Fear A.I. They Shouldn't' by David Brooks.
This is my critique of David Brooks opinion piece in the New York Times.
Tl;dr: Brooks believes that AI will never replace human intelligence but does not describe any testable capabilities that he predicts AI will never possess.
David Brooks arg... |
f3fc4e04-043c-4136-8c3a-6b934553de1e | trentmkelly/LessWrong-43k | LessWrong | Meetup : Australian-ish Online Hangout
Discussion article for the meetup : Australian-ish Online Hangout
WHEN: 20 August 2016 07:30:00PM (+1000)
WHERE: Canberra
It's back! Hopefully with fewer technical issues, since we'll try jitsi this time. I'll post the link on the night. https://www.facebook.com/events/9298009... |
77f4a737-4621-49aa-8d81-120f282ae161 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post169
I've just opened summer MATS applications (where I'll supervise people to write mech interp papers) I'd love to get applications from any readers who are interested! Apply here , due Feb 28 As part of this, I wrote up a list of research areas I'm currently excited about, and thoughts for promising directio... |
537c2fa5-c107-4d08-a4c1-c7bd37839df6 | trentmkelly/LessWrong-43k | LessWrong | How to export Android Chrome tabs to an HTML file in Linux (as of February 2023)
Let's say you have a few million tabs open in your mobile Chrome browser, because you never close anything, but now your browser is getting slow and laggy. You want to stick the URLs of those tabs somewhere for safekeeping so that you can... |
dd9a4860-5f29-452b-b577-9de3183cb9c7 | trentmkelly/LessWrong-43k | LessWrong | How do the different star-types in the universe (red dwarf, etc.) related to habitability for human-like life?
None |
2fc10f4c-5f55-4d16-8322-d155c5403fd8 | StampyAI/alignment-research-dataset/special_docs | Other | Policy desiderata in the development of machine superintelligence
Policy Desiderata for Superintelligent AI:
A Vector Field Approach
1
Nick Bostrom
†
, Allan Dafoe
†
, Carrick Flynn
†
[forthcoming in Liao, S.M. (ed.):
Ethics of Artificial Intelligence
(Oxford University Press, 2019)]
... |
93248547-5c26-48a3-8e05-05ad762e5a9e | trentmkelly/LessWrong-43k | LessWrong | Conventions and Confusing Continuity Conundrums
The next "How Not to be Stupid" may be a bit delayed for a couple of reasons.
First, there appears to be a certain unstated continuity assumption in the material I've been working from that would probably be relevant for the next posting. As I said in the intro post, I'... |
e508740a-e658-4bbd-b0ac-8ff56b44a3ac | trentmkelly/LessWrong-43k | LessWrong | LW November Diplomacy
The LW September Diplomacy game has come to an end. Congratulations to Prismattic, prase, and GuySrinivasan! It's time to see if we can start up another game.
If you would like to get in on a Diplomacy game with other LW users, please respond with a comment expressing your desire (between 1 and ... |
04445157-058b-4525-b239-65b60d4e7224 | trentmkelly/LessWrong-43k | LessWrong | Defending Functional Decision Theory
As I have been studying Functional Decision Theory (FDT) a lot recently, I have come across quite some counterarguments and general remarks that are worth rebutting and/or discussing in more detail. This post is an attempt to do just that. Most points have been discussed in other p... |
9de951e0-822f-46ae-b03a-2b9fc40d9eea | trentmkelly/LessWrong-43k | LessWrong | Naive comments on AGIlignment
Disclaimer: everything I know about AI I learned from reading stuff on the internet, mostly on this site.
1. Any self-modifying general intelligence cannot be bound by a utility function. The very nature of being able to self-modify means that the utility function is open to modificatio... |
4afc864d-d675-49f7-8500-2ba9651126ea | trentmkelly/LessWrong-43k | LessWrong | Please Press "Record"
Over the years, I've made very heavy use of open courseware/MOOCs. I'd estimate that I've covered about as much material in online lectures as I did in-person during a four year degree.
However, since college I've been frustrated by the general lack of online material for non-101 courses. Sites ... |
3486355d-8117-4b30-8507-e37da34ba578 | StampyAI/alignment-research-dataset/blogs | Blogs | Unraveling the evidence about violence among very early humans
*Click lower right to download or find on Apple Podcasts, Spotify, Stitcher, etc.*
As I've asked [Has Life Gotten Better?](https://www.cold-takes.com/has-life-gotten-better/), I've run into some intense debates about how violent early humans were.
* ... |
b9364317-5fa9-4499-a3d5-37d54a08b6d7 | trentmkelly/LessWrong-43k | LessWrong | Research agenda: Can transformers do system 2 thinking?
This is one of the posts that detail my research agenda, which tries to marry Deep Learning and chess to gain insight into current AI technologies. One of my main motivations for this research is trying to identify gaps aka unsolved problems in AI, because these ... |
514dd3d2-da27-4eba-9065-0333c09afb4b | trentmkelly/LessWrong-43k | LessWrong | Dice Decision Making
Epistemic status: I have been using dice on and off in my decision-making processes for a few years and am confident it has been a net positive for me (85%). I believe that experimenting with dice decisions would be beneficial for most people who struggle with decision-making (60%). I have a very ... |
6598036d-f6f8-4d39-8ccf-dd2c748d5a2c | trentmkelly/LessWrong-43k | LessWrong | Anthropic's Responsible Scaling Policy & Long-Term Benefit Trust
I'm delighted that Anthropic has formally committed to our responsible scaling policy. We're also sharing more detail about the Long-Term Benefit Trust, which is our attempt to fine-tune our corporate governance to address the unique challenges and long-... |
b480ae69-e9d0-4d1d-8b42-d95d2c2f33a8 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Frankfurt (including effective altruism presentation)
Discussion article for the meetup : Frankfurt (including effective altruism presentation)
WHEN: 27 October 2013 02:00:00PM (+0200)
WHERE: Frankfurt, Ginnheimer Landstraße
We have another meetup! The location is a private flat. Please contact me beforeha... |
99903a7c-aad1-4dc5-82f7-f615190b5841 | trentmkelly/LessWrong-43k | LessWrong | It's Not About The Nail
[This is hardly original; I’m documenting for my own sake since it took so long for me to understand. Cross-posted from Grand, Unified, Crazy.]
There’s an old saw, that when a women complains she wants sympathy, but when a man hears a complaint, he tries to solve the problem. This viral YouTub... |
f23a8b95-37cf-4b99-a134-c470e27f840c | trentmkelly/LessWrong-43k | LessWrong | What you can’t say to a sympathetic ear
Suppose we live in a society where it is strongly frowned upon to believe that an onion is a fruit. It is ok to disagree about what defines ‘fruit’, or what Allium varieties are onions. But none of this will get you off the hook—you had just better not suggest that an onions is ... |
63f01723-3c53-4885-aaae-9a4f6904ddf6 | trentmkelly/LessWrong-43k | LessWrong | Weekly LW Meetups: Salt Lake City, Melbourne, Atlanta, NYC, Cambridge
There are upcoming irregularly scheduled Less Wrong meetups in:
* Salt Lake City Meetup #2: 04 February 2012 03:00PM
* Atlanta: 04 February 2012 06:30PM
* First Brussels meetup: 11 February 2012 11:00AM
* Sydney Rationality meet-up No.2: 15 Feb... |
8766e1a0-5d9e-4129-a0f7-d451337942ee | trentmkelly/LessWrong-43k | LessWrong | Simulacra and Subjectivity
In Excerpts from a larger discussion about simulacra, following Baudrillard, Jessica Taylor and I laid out a model of simulacrum levels with something of a fall-from grace feel to the story:
1. First, words were used to maintain shared accounting. We described reality intersubjectively in ... |
9b14cb3a-0f7f-4085-a02a-3e879ef14db2 | trentmkelly/LessWrong-43k | LessWrong | Choose pain-free utilitarianism
Some of my friends are hedonic utilitarians, or close human approximations (people whose excuse to talk excitedly in bars and on the internet is sometimes hedonic utilitarianism). I am a preference utilitarian, so I would like to talk excitedly on the internet about how they are wrong.
... |
a16e607f-3abc-4d25-bdb7-6c1468adaf92 | trentmkelly/LessWrong-43k | LessWrong | Altruism and Vitalism Aren't Fellow Travelers
Summary: In his most recent post (“Altruism and Vitalism as Fellow Travelers”), Scott Alexander tries to reconcile EA-style altruism with Nietzschean “vitalism.” He claims that the philosophies argue for different goals only in extreme cases, while having similar goals in ... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.