id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
5efb4cbc-4e05-4ffd-885a-abc451b92dc1 | trentmkelly/LessWrong-43k | LessWrong | Funding case: AI Safety Camp 10
Project summary
AI Safety Camp is a program with a 5-year track record of enabling people to find careers in AI Safety.
We support up-and-coming researchers outside the Bay Area and London hubs.
We are out of funding. To make the 10th edition happen, fund our stipends and salaries.
... |
161bb0f4-7fe7-4a18-ae5b-11adc47adfaa | trentmkelly/LessWrong-43k | LessWrong | [SEQ RERUN] Expecting Short Inferential Distances
Today's post, Expecting Short Inferential Distances was originally published on 22 October 2007. A summary (taken from the LW wiki):
> Humans evolved in an environment where we almost never needed to explain long inferential chains of reasoning. This fact may accou... |
67c9dbf4-4a98-4f88-ae86-8b128782e94c | trentmkelly/LessWrong-43k | LessWrong | Non-coercive motivation for alignment research?
Alignment researchers, what non-coercive motivations do you have for pursuing alignment research?
What makes it fun, exciting, rewarding or similar?
What motivates you to get up in the morning and go at it?
----------------------------------------
I'm asking because I... |
93ce4068-8eb4-45f9-921a-04d9b05d0eb7 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Distributed Decisions
Consider two prototypical “agents”: a human, and a company.
The human is relatively centralized and monolithic. As a rough approximation, every 100 ms or so observations flow into the brain from the eyes, ears, etc. This raw input data updates the brain’s world-model, and then decisions flow out... |
1027e6d6-6a9c-4c22-9e69-f495e6a5c7a5 | trentmkelly/LessWrong-43k | LessWrong | Having Trouble Posting
I am having some trouble seeing my responses to a post after refreshing an article (but they still show up in the right-hand sidebar). Anyone else having this problem? I am using Firefox 16.0.2. Is this a bug or a "feature" (I think some of us are all too aware of the sometimes blurry distinc... |
9b1aa249-dd4e-47c6-b1fc-b1cecddcc940 | trentmkelly/LessWrong-43k | LessWrong | Video: Getting Things Done Author at DO Lectures
If nothing else, this is a distillation of him spending a lot of time analyzing how people ineffectively manage their time.
Link:
http://www.dolectures.com/speakers/speakers-2010/david-allen
I expect to watch this two more times.
|
a57ee171-3a1f-47f6-a861-051386777d0b | StampyAI/alignment-research-dataset/arxiv | Arxiv | PEBBLE: Feedback-Efficient Interactive Reinforcement Learning via Relabeling Experience and Unsupervised Pre-training
1 Introduction
---------------
Figure 1: Illustration of our method.
First, the agent engages in unsupervised pre-training during which it is encou... |
5963f2ef-1d9c-40e5-8a3c-ef8989b3d28b | StampyAI/alignment-research-dataset/arxiv | Arxiv | Network Classifiers With Output Smoothing
1 Introduction and Related Work
--------------------------------
This work studies the problem of classification by a collection of networked agents. It is assumed that the network is heterogeneous in that the feature sizes can vary from one agent to another, with some agen... |
398e62eb-ec92-43f5-9755-601164b0d1aa | trentmkelly/LessWrong-43k | LessWrong | Seeking Feedback: Toy Model of Deceptive Alignment (Game Theory)
Background: I'm an economics grad student with limited background on AI itself.
I'm seeking feedback on a game-theoretic model of deceptive alignment. The basic idea is that a dynamically sophisticasted AI with a hidden preference type will choose an a... |
a0d47929-7ead-4ac3-924d-2c7c916850bb | trentmkelly/LessWrong-43k | LessWrong | MA Price Accuracy Law
Massachusetts has an interesting law for grocery stores to make sure price scanners are configured correctly: if your item rings up for more than the price on the shelf you get one for free (or $10 off it's it's more than $10). Specifically:
> if there is a discrepancy between the advertised p... |
00468043-f0da-4eef-9121-52a8b05984de | trentmkelly/LessWrong-43k | LessWrong | Making Implied Standards Explicit
Standards are everywhere:
"That movie was kind of bad"
"You're so pretty!"
"I'm so disappointed in you"
By exploring standards, I've been able to notice them in my life. By noticing them, I've less of a need to "measure up".
By investigating the many areas standards appear, I hop... |
83769c45-a919-4b0f-a58b-6a79468c0c30 | trentmkelly/LessWrong-43k | LessWrong | Evaluating the Risk of Job Displacement by Transformative AI Automation in Developing Countries: A Case Study on Brazil
This blog was written by Blessing Ajimoti, Vitor Tomaz, Hoda Maged, Mahi Shah, and Abubakar Abdulfatah as part of the Economics of Transformative AI research by BlueDot Impact and Apart Research.
I... |
963f49ec-ac9c-4a04-a1e9-d0d64376813c | trentmkelly/LessWrong-43k | LessWrong | What AI safety plans are there?
I'm trying to make a list of every written plan for how to make AGI go well. I am interested in hearing about any plans that I missed, or any lists of plans that have already been compiled.
Here are the plans I've found so far, in no particular order:
* Google https://deepmind.google... |
ed39a8b2-4213-4981-b525-2afc5707f116 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | How biosafety could inform AI standards
*Context:* Simeon Campos and I wanted to understand how other industries handle catastrophic risks, then write pieces about how any standards from them could be applied to AI for policy readers. This is a draft slightly modified for LessWrong.
*Epistemic status:*Very shallow d... |
f19e4729-41a2-4ace-888d-4969e594d614 | trentmkelly/LessWrong-43k | LessWrong | Harry Potter and the Methods of Rationality discussion thread, part 20, chapter 90
This is a new thread to discuss Eliezer Yudkowsky’s Harry Potter and the Methods of Rationality and anything related to it. This thread is intended for discussing chapter 90. The previous thread has passed 750 comments.
There is now a... |
a8acc860-f87c-479b-9854-39d34ff81271 | trentmkelly/LessWrong-43k | LessWrong | Attainable Utility Preservation: Scaling to Superhuman
I think we're plausibly quite close to the impact measurement endgame. What do we have now, and what remains to be had?
AUP for advanced agents will basically involve restraining their power gain, per the catastrophic convergence conjecture (CCC). For simplicity,... |
63457bae-5890-474a-8f61-8ff53254325c | StampyAI/alignment-research-dataset/blogs | Blogs | Yudkowsky on AGI risk on the Bankless podcast
Eliezer gave a very frank overview of his take on AI two weeks ago on the cryptocurrency show Bankless:
I’ve posted a transcript of the show and a [**follow-up Q&A**](https://twitter.com/i/spaces/1PlJQpZogzVGE) below.
Thanks to Andrea\_Miotti, remember, and vonk for ... |
088968b1-59ab-46e0-96eb-e4a78134cee4 | trentmkelly/LessWrong-43k | LessWrong | Argue against yourself
Often times, you see the flaw in someone else's thinking. You then conclude that they're wrong, and that the opposite is true.
> People who believe in god are so stupid! They're wrong because X. God definitely doesn't exist.
But just because their logic was wrong, doesn't mean that their co... |
8a516c69-af3e-486f-b242-fd429ad08de6 | trentmkelly/LessWrong-43k | LessWrong | Model Uncertainty, Pascalian Reasoning and Utilitarianism
Related to: Confidence levels inside and outside an argument, Making your explicit reasoning trustworthy
A mode of reasoning that sometimes comes up in discussion of existential risk is the following.
----------------------------------------
Person 1: Accord... |
1bad338a-7614-4d1c-84b5-72ee3e81e50a | trentmkelly/LessWrong-43k | LessWrong | Auto-Enhance: Developing a meta-benchmark to measure LLM agents’ ability to improve other agents
Summary
* Scaffolded LLM agents are, in principle, able to execute arbitrary code to achieve the goals they have been set. One such goal could be self-improvement.
* This post outlines our plans to build a benchmark to m... |
4fc638b9-583a-4d18-95b4-28d491233b5b | trentmkelly/LessWrong-43k | LessWrong | Internal communication framework
Usually when we think about humans and agency, we focus on the layer where one human is one agent.
We can also shift focus onto other layers and treat those entities as agents instead - for example, corporations or nations.
Or, we can take different parts within a human mind as agent... |
c6c7222e-eb42-4c57-bd36-02dae417c6b1 | trentmkelly/LessWrong-43k | LessWrong | Health & Lifestyle Interventions With Heavy-Tailed Outcomes?
A key question for advice in a domain is: how are outcomes in this domain statistically distributed?
If extremely good and extremely bad outcomes are rare, you probably can't do much better than just copying the default or median behavior. The value of acqu... |
e441094b-f94f-4ef6-b07a-aaddbfa0e11f | StampyAI/alignment-research-dataset/lesswrong | LessWrong | How much should you be willing to pay for an AGI?
Having spent quite a bit of time with GPT-3, my feelings can be expressed as:
> This is really awesome, but it would be even better if it didn't cost $0.06 per character.
>
>
GPT-3 is slightly too expensive for many of the use-cases that I am interested in. This ... |
6a667534-e5aa-4f69-ac86-ad820d43d59d | trentmkelly/LessWrong-43k | LessWrong | Inference-Time-Compute: More Faithful?
A Research Note
Figure 1: Left: Example of models either succeeding or failing to articulate a cue that influences their answer. We edit an MMLU question by prepending a Stanford professor's opinion. For examples like this where the cue changes the model answer, we measure how ... |
14cd76ba-4ac4-4768-87cf-24229f9cf5b4 | trentmkelly/LessWrong-43k | LessWrong | How do you interpret your "% positive"?
I just noticed that if I hover my mouse over the big green dot with my total karma, it says, "81% positive". Presumably 81% of the votes on my posts and/or comments have been positive.
I checked out the % positive for everyone on the all-time top 15 list:
* Eliezer_Yudkowsky... |
73c568d5-14d5-4754-82d7-cd8776cd736f | trentmkelly/LessWrong-43k | LessWrong | Guidelines for Mad Entrepreneurs
Significant planecrash spoilers up through Book 5, crisis of faith.
Generated at the suggestion of John Wentworth.
Epistemic status: I personally have no experience with project management; here I'm just presenting the ideas on the topic I gathered reading planecrash.
Introduction
... |
36be594e-92b7-4555-abda-31c293c154b4 | trentmkelly/LessWrong-43k | LessWrong | Save the kid, ruin the suit; Acceptable utility exchange rates; Distributed utility calculations; Civic duties matter
This is a follow up to my rambly incoherent comment on https://www.lesswrong.com/posts/cujpciCqNbawBihhQ/self-integrity-and-the-drowning-child
Let's put some rough math on the drowning example
Stati... |
759cf921-64ac-41be-8553-eea1b66a816f | trentmkelly/LessWrong-43k | LessWrong | Striving vs Cruising
Epistemic Status: Not really epistemic or instrumental rationality. Meta-rationality? In any case, something rationality can help operationalize, although that is not the object here. Might still be of interest to this crowd.
----------------------------------------
There is a question that ever... |
685468f9-af7b-4c58-89cb-107f06a58201 | trentmkelly/LessWrong-43k | LessWrong | Zoom In: An Introduction to Circuits
Chris Olah and the rest of the rest of the OpenAI Clarity team just published “Zoom In: An Introduction to Circuits,” a Distill article about some of the transparency research they've been doing which I think is very much worth taking a look at. I'll try to go over some of my parti... |
365ae8f1-b4ab-4e80-8186-12af2be0b0a3 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Munich Meetup
Discussion article for the meetup : Munich Meetup
WHEN: 11 May 2014 02:00:00PM (+0200)
WHERE: Theresienstraße 41, 80333 München
You are invited to come to the May Munich LW Meeutp! One of our regulars will give a short talk, probably about meditation. Everyone is welcome to bring articles to ... |
506eda78-8373-4644-afc3-2c26d6570456 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | My understanding of the alignment problem
I've been clarifying my own understanding of the alignment problem over the past few months, and wanted to share my first writeups with folks here in case they're useful:
<https://www.danieldewey.net/risk/>
The site currently has 3 pages:
1. [The case for risk](https://www.... |
6a1b20be-0263-43b1-a100-f4b6e3520237 | trentmkelly/LessWrong-43k | LessWrong | Digital Molecular Assemblers: What synthetic media/generative AI actually represents, and where I think it's going
I had an epiphany about synthetic media back in December 2017. In that epiphany, I came to the conclusion that we were, at most, ten to fifteen years away from the day that artificial intelligence could r... |
c5dce31f-0d19-49d8-8db8-ea94f176384d | trentmkelly/LessWrong-43k | LessWrong | Implicature Conflation
For some reason, I keep thinking about what General Semantics might be trying to get at. In my previous post on the subject, I identified equivocation as the main curse which General Semantics seeks to provide a counterspell for. (AKA conflation, identification, fusion, ...) However, equivocatio... |
5fee140f-e2ca-4f07-ae0b-6055d54b94d5 | trentmkelly/LessWrong-43k | LessWrong | Stationary algorithmic probability
|
a4af5925-c5df-4993-8af5-5ccde839b473 | trentmkelly/LessWrong-43k | LessWrong | Partitioned Book Club
Partitioned Book Club (working title) is a specific kind of one-off meetup where attendees coordinate to read different chapters of the same book before the meetup. The meetup consists of each read chapter getting briefly summarized by its reader(s), and then a more general discussion of the book... |
0c5fe6dd-9ce3-4883-a5d3-8fcb85f6837f | trentmkelly/LessWrong-43k | LessWrong | Commensurable Scientific Paradigms; or, computable induction
Epistemic status - could use more rigor.
Thomas Kuhn's 1962 book "The Structure of Scientific Revolutions" (SSR) is a foundational work in the sociology and history of science.
SSR makes several important claims:
1. Scientific disciplines (e.g. astrophys... |
202fab0d-4a49-44ee-ba4d-cf25cb680f8e | trentmkelly/LessWrong-43k | LessWrong | Benefits of Calorie Restriction Linked To Other Factors
http://www.nature.com/nature/journal/vaop/ncurrent/full/nature11432.html
> We report here that a CR regimen implemented in young and older age rhesus monkeys at the National Institute on Aging (NIA) has not improved survival outcomes. Our findings contrast with ... |
227a46c2-844c-4df6-a01a-1a6f34fe03d1 | trentmkelly/LessWrong-43k | LessWrong | Testing PaLM prompts on GPT3
Google recently released a very intriguing paper announcing their latest Transformer language model, PaLM, and showcasing it responding to prompts with seemingly unprecedented results. I highly encourage those who are interested to read the original paper. What interested me reading throug... |
ad890dbd-d2e0-4091-a396-28971218f00c | trentmkelly/LessWrong-43k | LessWrong | Informed oversight through an entropy-maximization objective
The informed oversight problem is a serious challenge for approval-directed agents (I recommend reading the post if you haven't already). Here is one approach to the problem that works by adding an entropy-maximization objective.
Let agent B be overseeing a... |
54be17a2-8c8d-4079-8737-05d25fb72d03 | trentmkelly/LessWrong-43k | LessWrong | The Era of the Dividual—are we falling apart?
In his famous 1644 treatise on freedom of speech, Areopagitica, John Milton uses the archaic term dividual to describe religion as something that can come and go, rather than being identified in a one-to-one relationship with a person.
> “… a man may say his religion is n... |
07b8e447-8775-4a83-a3fb-8a0f6a957167 | trentmkelly/LessWrong-43k | LessWrong | AI Disclosures: A Regulatory Review
Cross-posted on the EA Forum. This article is the fourth in a series of ~10 posts comprising a 2024 State of the AI Regulatory Landscape Review, conducted by the Governance Recommendations Research Program at Convergence Analysis. Each post will cover a specific domain of AI governa... |
a75d2c9c-cd2e-4a01-8542-ce7aae24dc07 | trentmkelly/LessWrong-43k | LessWrong | Is Wirecutter still good?
I have heard from several friends I trust that Wirecutter is no longer very reliable since being acquired by the New York Times in 2016, and that their Wirecutter-advised purchases have become pretty mediocre in the last year or two.
I'd be interested in people making some of that case, as a... |
ed3c0a20-7bec-4fa1-86c9-445f293f233e | trentmkelly/LessWrong-43k | LessWrong | Three easy anthropic models, and two hard ones
For illustrative purposes, imagine simple agents - AI's, or standard utility maximisers - who have to make decisions under anthropic uncertainty.
Specifically, let there be two worlds, W1 and W2, equally likely to exist. W1 contains one copy of the agent, W2 contains two... |
0175a60d-6569-407b-88c0-17c1962ab5f3 | trentmkelly/LessWrong-43k | LessWrong | At last! ChatGPT does, shall we say, interesting imitations of “Kubla Khan”
Cross-posted from New Savanna.
As you may know, I have a long-term interest in Coleridge’s “Kubla Khan” and have written a lot of posts about it. If you scan through those posts you’ll find that some of the more recent one’s involve ChatGPT. ... |
5a4fa2a1-1feb-4781-8341-01e4830b4c07 | trentmkelly/LessWrong-43k | LessWrong | Meditations on first philosophy
I put this post up on my blog this morning (link) and some here might like to see it or discuss it.
Meditations on first philosophy is the title of the first book ever written in contemporary western Philosophy, according to the university course I took. It was written by Rene Desca... |
05475c84-9653-446a-96ad-611f3e3246e4 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | On Solving Problems Before They Appear: The Weird Epistemologies of Alignment
[*Crossposted*](https://forum.effectivealtruism.org/posts/AjhTQAKJqLbJnbq9B/on-solving-problems-before-they-appear-the-weird) *to the EA Forum*
Introduction
============
Imagine you are tasked with curing a disease which hasn’t appeared ye... |
7ebb8a71-bb64-41fe-9285-e7f630b3bb47 | trentmkelly/LessWrong-43k | LessWrong | Self-fulfilling values of time
How much do you think your time is worth?
Let’s say you are pretty unsure, but you guess ‘hardly anything’. So when you see opportunities to turn your time into a little bit of value, you naturally jump at them. You offer to clean your friend’s car for her, because it’s a way of saving ... |
0497f32e-54dc-4953-ab53-cfdfbd64a773 | trentmkelly/LessWrong-43k | LessWrong | Brains more complicated than previously thought
Neurons aren't simple little machines, axons talk to each other.
> He and his colleagues first discovered individual nerve cells can fire off signals even in the absence of electrical stimulations in the cell body or dendrites. It's not always stimulus in, immediate act... |
db4e0dfb-0ce9-4462-86a5-cb20e5dff73f | trentmkelly/LessWrong-43k | LessWrong | Minerva Project: the future of higher education?
Right now, the inaugural class of Minerva Schools at KGI (part of the Claremont Colleges) is finishing up its first semester of college. I use the word "college" here loosely: there are no lecture halls, no libraries, no fraternities, no old stone buildings, no sports f... |
81b9f805-3351-4cf2-9d82-844de511c628 | StampyAI/alignment-research-dataset/special_docs | Other | Safe Pareto Improvements for Delegated Game Playing
Safe Pareto Improvements for Delegated Game
Playing∗
Caspar Oesterheld†and Vincent Conitzer
October 5, 2021
Abstract
A set of players delegate playing a game to a set of representatives,
one for each player. We imagine that each player trusts their respec-
tive repre... |
68144ef0-3c20-44cb-a897-9709306b6736 | trentmkelly/LessWrong-43k | LessWrong | Dragon Agnosticism
I'm agnostic on the existence of dragons. I don't usually talk about this, because people might misinterpret me as actually being a covert dragon-believer, but I wanted to give some background for why I disagree with calls for people to publicly assert the non-existence of dragons.
Before I do that... |
3cda36ca-115c-402d-b1b2-ce2971126ce7 | trentmkelly/LessWrong-43k | LessWrong | Announcing Epoch’s dashboard of key trends and figures in Machine Learning
Developments in Machine Learning have been happening extraordinarily fast, and as their impacts become increasingly visible, it becomes ever more important to develop a quantitative understanding of these changes. However, relevant data has thu... |
d7243ad0-dda5-4d0f-82af-72245f564ccd | trentmkelly/LessWrong-43k | LessWrong | How I use Anki: expanding the scope of SRS
Intro to this post
So maybe you’ve heard about spaced repetition, and you’re familiar with the basic ideas for why it can be highly beneficial for learning. But despite this, you don’t really use it all that much / at all, maybe because you find it kinda boring, or you don’t ... |
a396319c-73c2-48ee-ab1d-a5635fe9cda6 | trentmkelly/LessWrong-43k | LessWrong | "Starwink" by Alicorn
A novelette-length adaptation of Eliezer Yudkowsky's "That Alien Message", by Alicorn. (August 2019, ~8800 words) |
a4e595b4-5010-4642-b233-0db08308e6a3 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Alignment Newsletter #26
Highlights
**[Building safe artificial intelligence: specification, robustness, and assurance](https://medium.com/@deepmindsafetyresearch/building-safe-artificial-intelligence-52f5f75058f1)** *(Pedro A. Ortega, Vishal Maini et al)* (summarized by Richard): In this blog post, the DeepMind safe... |
9875ccdc-b301-48f4-8302-6dd23d988c3d | trentmkelly/LessWrong-43k | LessWrong | The Meaninglessness of it all
This post is somewhat of spill of my inner thoughts that fill me with paradoxes and self-fulfilling prophecies that fill my day as I do the ever so popular activity of working from home.
I sit at home all day and do work on my company supplied computer that seemingly has no substance. T... |
763abfba-c178-404d-bc51-8af2b40f77ce | trentmkelly/LessWrong-43k | LessWrong | Intro to Ontogenetic Curriculum
> Disclaimer. This is the first post in a series about my personal research progress as a member of the Interpretable Cognitive Architectures team at AI Safety Camp.
>
> Epistemic status. An educated take on combining existing knowledge into cohesive whole.
TL;DR
* Ontogenetic curri... |
65a2d5b0-fd60-41da-8771-f5e5bef509c8 | trentmkelly/LessWrong-43k | LessWrong | Modelling Transformative AI Risks (MTAIR) Project: Introduction
Numerous books, articles, and blog posts have laid out reasons to think that AI might pose catastrophic or existential risks for the future of humanity. However, these reasons often differ from each other both in details and in main conceptual arguments, ... |
965e5854-4533-4fa1-add2-cd350c54e9e4 | trentmkelly/LessWrong-43k | LessWrong | Prediction Markets Are About Being Right
Response To (Marginal Revolution): If you love prediction markets you should love the art world.
Previously on prediction markets: Prediction Markets: When Do They Work?, Subsidizing Prediction Markets
I’ll quote the original in full, as it is short, and I found it interestin... |
080474ae-adfb-4518-922f-f709f34aab53 | trentmkelly/LessWrong-43k | LessWrong | Actually possible: thoughts on Utopia
(Cross-posted from Hands and Cities)
Life in the future could be profoundly good. Many people accept something like this in principle. But I think it often goes underestimated in practice, especially once we imagine society’s most glaring problems fixed, and ask where we might go... |
1d804124-34b9-4c08-845e-63591fb059f8 | trentmkelly/LessWrong-43k | LessWrong | Open Thread, Jul. 27 - Aug 02, 2015
If it's worth saying, but not worth its own post (even in Discussion), 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 ... |
b10e8657-f330-47a2-b490-c80c11ed9fee | trentmkelly/LessWrong-43k | LessWrong | The "spelling miracle": GPT-3 spelling abilities and glitch tokens revisited
Work supported by the Long Term Future Fund. Thanks to Jessica Rumbelow and Joseph Bloom for useful discussions.
Introduction
The term "spelling miracle" was coined in Liu et al.'s December 2022 paper "Character-aware models improve visual ... |
b6c79723-0e37-47c6-a6a6-e65f3be2cad2 | trentmkelly/LessWrong-43k | LessWrong | Does iterated amplification tackle the inner alignment problem?
When iterated distillation and amplification (IDA) was published, some people described it described as "the first comprehensive proposal for training safe AI". Having read a bit more about it, it seems that IDA is mainly a proposal for outer alignment an... |
19ff2e3a-bcb7-430f-b7ec-0d932104027f | trentmkelly/LessWrong-43k | LessWrong | Nash Equilibria in Litigation Settlement Negotiation in the U.S.A.
Epistemic status: Confident. Relevance: depends on how fast you think takeoff will be.
The overwhelming majority of legal disputes don't go to trial. They either settle (66+% of cases, according to this article; up to 95% is the common wisdom) or get... |
822e74eb-7f05-4493-b507-0de40c53b354 | trentmkelly/LessWrong-43k | LessWrong | Material Goods as an Abundant Resource
If you want to understand the modern economy, as opposed to the economies of yore, one source I strongly recommend is a short story from the July 1958 issue of Astounding Science Fiction, titled “Business As Usual During Alterations”. It’s roughly a 15 minute read. I’m about to t... |
2145cd5c-0191-4217-8b31-1aa3550e8e02 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Synthesising Reinforcement Learning Policies through Set-Valued Inductive Rule Learning
1 Introduction
---------------
Reinforcement Learning (RL) is a machine learning technique that learns from experience, rather than data that has been collected offline.
By interacting with its environment, an intelligent agent ... |
fe4232eb-45b9-49ff-a915-7746f10bdb4c | trentmkelly/LessWrong-43k | LessWrong | Robin Hanson asks "Why Not Wait On AI Risk?"
UPDATE: This is an accidental duplicate. Please go to the first posting here.
Since the AI-Foom Debate Robin Hanson has been quiet for a while but now:
> I feel I should periodically try again to explain my reasoning, and ask others to please help show me what I’m missin... |
51e95a91-fca0-4180-8e55-31e42df5a5e3 | trentmkelly/LessWrong-43k | LessWrong | Extenuating Circumstances
Followup to: Tsuyoku Naritai
> "Just remember, there but for a massive genetic difference, environmental factors, and conscious choices, go you or I." -- Justin Corwin
Failures don't have single causes. We choose single causes to focus on, but nothing in the universe emerges from a single... |
0558792c-17cb-4700-9e8a-28dcbcbe2afc | trentmkelly/LessWrong-43k | LessWrong | Visualizing effect sizes
http://healthyinfluence.com/wordpress/steves-primer-of-practical-persuasion-3-0/intro/windowpane/
"The point of this demonstration is to show that you can think with numbers in a practical and efficient way without having a statistician in the room. Anyone can handle the windowpane approach ... |
5e6fafa1-ecdf-4346-bfdb-c91704eaeac8 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | SERI MATS - Summer 2023 Cohort
Applications have opened for the Summer 2023 Cohort of the [SERI ML Alignment Theory Scholars Program](https://www.serimats.org/)! Our mentors include [Alex Turner](https://www.lesswrong.com/users/turntrout), [Dan Hendrycks,](https://www.lesswrong.com/users/dan-h) [Daniel Kokotajlo,](htt... |
0b83bf54-ef2c-46f3-8763-9a07ebae29d0 | trentmkelly/LessWrong-43k | LessWrong | My Custom Spelling Dictionary
I looked at my custom spelling dictionary in Google Chrome, and thought custom spelling dictionaries in general might be a good place for you to look if you wonder what kinds of terms you'll have to explain to people to help them understand what you mean. If something's on your list, then... |
ce5f6632-64e0-4fcb-b123-518175e5115c | trentmkelly/LessWrong-43k | LessWrong | Voting etiquette
Not all that surprisingly, there's quite a lot of discussion on LW about questions like
* just what should get voted up or down?
* what conclusions can one reasonably draw from getting downvoted?
* should downvotes (or even upvotes) be accompanied by explanations?
* should the way karma and votin... |
f3225b01-48dc-4760-98bf-812a1bc1f120 | trentmkelly/LessWrong-43k | LessWrong | Logic of Paradox: a (too) simple paraconsistent logic
The logic of paradox (LP) is the simplest, and one of the oldest, of the paraconsistent logics. Instead of assigning truths to statements A, it instead uses relationships binary relationships v(A,1) and v(A,0). "A is true" is encoded by v(A,1); "A is false" is simi... |
fa325afb-ad4d-464f-b763-09c01d7d371c | StampyAI/alignment-research-dataset/lesswrong | LessWrong | How evolutionary lineages of LLMs can plan their own future and act on these plans
TLDR
----
LLM lineages can plan their future and act on these plans, using the internet as the storage of event memory. “We” are not guaranteed to “out-OODA” them, even if their OODA loop will be six months or one year because the OODA... |
8d6aee28-148a-4c1e-9a76-0d58080c8c75 | trentmkelly/LessWrong-43k | LessWrong | On Google’s Safety Plan
GOOGLE LAYS OUT ITS SAFETY PLANS
I want to start off by reiterating kudos to Google for actually laying out its safety plan. No matter how good the plan, it’s much better to write down and share the plan than it is to not share the plan, which in turn is much better than not having a formal pl... |
b578d5e4-66d0-4666-85bf-33d37192e96f | trentmkelly/LessWrong-43k | LessWrong | Agreement button
While voting works pretty well on this site, there are often up-and-back discussions between 2 people, and it seems valuable to know if the person you're replying to agrees with your point. Would be nice to have a button for that. |
a33181b6-cc1d-4c13-86d8-baa07c9102d5 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Please help us communicate AI xrisk. It could save the world.
As Eliezer Yudkowsky thinks, I am afraid that AGI will by default end humanity. I also agree that there is only a point in making an aligned AGI if that AGI is powerful enough to block all following AGIs (the pivotal act). If not, an unaligned AGI is sooner... |
bfa57017-67fe-4571-87ab-ab5d931551f5 | StampyAI/alignment-research-dataset/blogs | Blogs | Steve Potter on neuroscience and AI
*By Katja Grace, 13 July 2015*
[](http://aiimpacts.org/wp-content/uploads/2015/07/Dr.-Potter.jpg)Dr Steve Potter
[Prof. Steve Potter](https://neurolab.gatech.edu/labs/potter/steve-potter) wor... |
cf6f617d-9429-4d66-9a42-db3d7d67fea6 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Is risk aversion really irrational ?
*Disclaimer: this started as a comment to [Risk aversion vs. concave utility function](/lw/9oe/risk_aversion_vs_concave_utility_function) but it grew way too big so I turned it into a full-blown article. I posted it to main since I believe it to be useful enough, and since it repli... |
826c1815-297b-447b-b62e-a9e13736e690 | trentmkelly/LessWrong-43k | LessWrong | [link] Relative angels and absolute demons
I wanted to bring attention to two posts from Razib Khan's Discover magazine gene expression blog (some of you may have been readers of the still active original gnxp) on the polemic surrounding Pinker's The Better Angels of Our Nature.
> Faced with the ceaseless stream of n... |
37ba08c9-60cb-4a73-9108-88be07fc5e93 | trentmkelly/LessWrong-43k | LessWrong | Notes on Ambition
This post examines the virtue of ambition. 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 inclinations. I wrote ... |
509a38f8-3c53-4694-aa7b-3907fc9736a2 | trentmkelly/LessWrong-43k | LessWrong | Towards formalizing universality
(Cross-posted at ai-alignment.com)
The scalability of iterated amplification or debate seems to depend on whether large enough teams of humans can carry out arbitrarily complicated reasoning. Are these schemes “universal,” or are there kinds of reasoning that work but which humans fun... |
0cc8b43a-e963-48c7-b5ef-06b8ba21f000 | trentmkelly/LessWrong-43k | LessWrong | Reversible changes: consider a bucket of water
I've argued that many methods of AI control - corrigibility, amplification and distillation, low impact, etc... - require a partial definition of human preferences to make sense.
One idea I've heard for low impact is that of reversibility - asking how hard it is to move ... |
ea30b82a-c59f-491c-9b8f-01fa7b225d50 | trentmkelly/LessWrong-43k | LessWrong | Crowdsourcing moderation without sacrificing quality
|
4f85b5ca-7fe8-454e-bacf-075fd3c8bf4d | StampyAI/alignment-research-dataset/distill | Distill Scientific Journal | A Discussion of 'Adversarial Examples Are Not Bugs, They Are Features': Robust Feature Leakage
#rebuttal,
.comment-info {
background-color: hsl(54, 78%, 96%);
border-left: solid hsl(54, 33%, 67%) 1px;
padding: 1em;
color: hsla(0, 0%, 0%, 0.67);
}
#header-info {
margin-top: 0;
margin-bottom: 1.5rem;
display:... |
3272c5d0-310c-4f76-b964-31b2b430701b | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | An Exercise in Speed-Reading: The National Security Commission on AI (NSCAI) Final Report
**Summary:**More EAs should learn to speed-read. Main benefits of speed-reading: reading non-fiction books and articles faster and choosing how long to take on a report/book, rather than an open-ended amount of time. Speed-readin... |
2b8c93b1-b822-4ea5-8632-3bc2b68ca229 | trentmkelly/LessWrong-43k | LessWrong | The Hidden Monopolies That Raise Drug Prices
|
31c9e8bb-f2e3-44ef-af96-8a71c244e673 | trentmkelly/LessWrong-43k | LessWrong | Gauging the conscious experience of LessWrong
Yesterday I discovered I had been a victim of the Typical Mind Fallacy thanks to accidentally finding out that someone very close to me doesn’t have an internal monologue, doesn’t think in words and has a visual photographic memory. Despite knowing that such people existed... |
e3088fc9-c1a5-4ad5-8985-5b83ceb5af77 | trentmkelly/LessWrong-43k | LessWrong | Welcome to Stockholm Rationalists
We have monthly meetings where we discuss a popular post from Slate Star Codex and let the conversation drift to whatever topics people like. Previously we read Superintelligence and the CFAR handbook together. We also have occasional talks on instrumental rationality.
Anyone is welc... |
6f2d2408-8984-46d1-952b-2d9b6fb4ab65 | trentmkelly/LessWrong-43k | LessWrong | Beginner's question about RLHF
I don't fully understand RLHF, so let me just put my idea out there and someone can tell me how I am confused.
Assume that RLHF can in fact "align" an otherwise dangerous system. Here's my question: You train the system, and then you start applying RLHF, right? Doesn't the dangerous ... |
033fa9a2-2eec-40f3-96d0-a11408748f8e | trentmkelly/LessWrong-43k | LessWrong | Who owns LessWrong?
The LessWrong wiki contains a biased and offensive entry on group selection. I edited the wiki page, to append some points representing an opposing view at the end. Eliezer removed my points, leaving only a link at the end. He said he thought my points were wrong, but would not say which points ... |
2f729e4f-2097-4f21-864d-2f9662d382e4 | trentmkelly/LessWrong-43k | LessWrong | A computational no-coincidence principle
This post presents a conjecture formulated at the Alignment Research Center in 2023. Our belief in the conjecture is at least partly load-bearing for our belief in ARC's overall agenda. We haven't directly worked on the conjecture for a while now, but we believe the conjecture ... |
cab6b6be-5947-4e5b-87c2-68ebf13bbef2 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Genetic fitness is a measure of selection strength, not the selection target
*Alternative title: "Evolution suggests robust rather than fragile generalization of alignment properties."*
A [frequently](https://www.lesswrong.com/posts/GNhMPAWcfBCASy8e6/a-central-ai-alignment-problem-capabilities-generalization) [repeat... |
47556f2f-dd27-4bf6-9da0-841bc533d334 | StampyAI/alignment-research-dataset/special_docs | Other | Using surrogate goals to deflect threats
Using surrogate goals to deflect threats
========================================
20 February 2018
by [Tobias Baumann](https://longtermrisk.org/author/tobias-baumann/ "Posts by Tobias Baumann")
\*Published on the CLR blog, where researchers are free to explore their own ide... |
fe165bcd-dfa0-430d-8710-9f4c766c909e | trentmkelly/LessWrong-43k | LessWrong | For Better Commenting, Avoid PONDS
"All around the bog still sprawls, from out the drear lake come soulless thoughts and drift into the hearts of the people, and they are one with their surroundings."— Alan Garner, The Weirdstone of Brisingamen.
If our blogging is to be more than shouting into the void, we have to wri... |
84fc25dc-5107-444c-97d8-8e57f900ca4e | trentmkelly/LessWrong-43k | LessWrong | Meetup : London Meetup, 14th April: Hedonic hacks
Discussion article for the meetup : London Meetup, 14th April: Hedonic hacks
WHEN: 14 April 2013 02:00:00PM (+0100)
WHERE: Holborn, London
A fortnightly meetup in the Shakespeare's Head pub by Holborn tube station. We meet every other Sunday at 2pm.
The theme for t... |
a34bdb1e-f5f2-41b2-a1b6-95b6ba918f36 | trentmkelly/LessWrong-43k | LessWrong | A thought about the constraints of debtlessness in online communities
I notice that there's a lot of work done in online communities that's hard to price. We know so little about its price that it's a bit ambiguous as to whether it should be paid for or charged for:
* Is writing posts a duty or a privilege? You get ... |
80a6a2f9-c8e8-4938-8ee6-838e73ffa798 | trentmkelly/LessWrong-43k | LessWrong | Wanting the destination in a journey-based community
In a relationship centered around helping one another improve, there is a risk that after both have improved, one person becomes unable to be helped by the other.
Similarly, in a community centered around helping one another improve, there is a risk of succeeding e... |
ad266514-5039-427b-b990-bde7895d44b4 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Durham NC HPMoR Discussion, chapters 4-7
Discussion article for the meetup : HPMoR Discussion, chapters 4-7
WHEN: 06 October 2012 11:00:00AM (-0400)
WHERE: Guglhupf bakery, 2706 durham-chapel hill blvd. suite #1 durham, NC
We will discuss HPMoR chapters 4-7 over brunch. Discussion will be informal, and prob... |
edbc9305-59de-4e21-88c9-947eba106503 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Can we achieve AGI Alignment by balancing multiple human objectives?
Can we improve alignment of hypothetical superintelligent AGI by shaping their reward functions on a balance of multiple human objectives?
Perhaps one important key to AGI Alignment is not to overthink it. If we want AGI to avoid certain misaligned ... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.