id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
c7a3f8c3-6105-4915-a589-03f02fe7f17e | trentmkelly/LessWrong-43k | LessWrong | Rational Justice
I'm interested in how courts and juries might use rational techniques to arrive at correct decisions on guilt.
In a complex case, it would seem to sensible to assess each component of the prosecution and defence case, and estimate the relative likelihood. If the prosecution case is (say) 100 times mo... |
94895bd8-91f3-43b1-b7a4-66e820462856 | trentmkelly/LessWrong-43k | LessWrong | Concept Safety: The problem of alien concepts
In the previous post in this series, I talked about how one might get an AI to have similar concepts as humans. However, one would intuitively assume that a superintelligent AI might eventually develop the capability to entertain far more sophisticated concepts than humans... |
7556056e-36c5-4f19-8282-038f6564d024 | trentmkelly/LessWrong-43k | LessWrong | Sunday August 2, 12pm (PDT) — talks by jimrandomh, johnswenthworth, Daniel Filan, Jacobian
This Sunday at 12pm (PDT), we're running another session of "lightning talks" by curated LessWrong authors (see here for previous weeks' transcripts).
* Each talk will be 3-5 minutes followed by discussion. Afterwards, we'll h... |
b6c13957-2164-4ca7-bb5a-5eb0cca8813e | trentmkelly/LessWrong-43k | LessWrong | Rethinking Batch Normalization
Yesterday we saw a glimpse into the inner workings of batch normalization, a popular technique in the field of deep learning. Given that the effectiveness of batch normalization has been demonstrated beyond any reasonable doubt, it may come as a surprise that researchers don't really kno... |
7ef1047f-305d-4ee1-854d-0c9348a09cc1 | trentmkelly/LessWrong-43k | LessWrong | Against naming things, and so on
Recent discussion on naming concepts mostly focuses on arguments in favor, noting only a few caveats, as LW user Conor Moreton in Why and How to Name Things:
> What you lose by the proliferation of jargon is ease-of-entry and cross-cultural intelligibility and hard-drive space in the ... |
1f835170-7360-4b1f-9613-42df773b9f30 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "(An idea I had while responding to this quotes thread)
"Correlation does not imply causation" is bandied around inexpertly and inappropriately all over the internet. Lots of us hate this.
But get this: the phrase, and the most obvious follow-up phrases like "what does imply causation?" are not high-competition search ... |
4c1591be-553f-4aaa-a6f4-d6e1b5c15d2c | trentmkelly/LessWrong-43k | LessWrong | What's the largest sunk cost you let go?
None |
a06c618b-9489-4542-8b29-839d549d77c7 | trentmkelly/LessWrong-43k | LessWrong | The Principled Intelligence Hypothesis
I have been reading the thought provoking Elephant in the Brain, and will probably have more to say on it later. But if I understand correctly, a dominant theory of how humans came to be so smart is that they have been in an endless cat and mouse game with themselves, making norm... |
e04aa4eb-55e7-4f8e-a6b9-41891199dfe6 | trentmkelly/LessWrong-43k | LessWrong | Focusing your impact on short vs long TAI timelines
Summary/Key Points
I compare considerations for prioritizing impact in short vs. long transformative AI timeline scenarios. Though lots of relevant work seems timelines-agnostic, this analysis is primarily intended for work whose impact is more sensitive to AI timel... |
193fecfa-cddc-47a1-b40b-4fdf0db27ff6 | trentmkelly/LessWrong-43k | LessWrong | Tax-Optimized Risk in Portfolio Allocation
When you're subject to capital gains taxation, the government shares in some of the upside, but when you have capital losses, the government shares in the downside too[1]. Because of this, the actual risk (and reward) of any given portfolio is lower than it seems. To countera... |
62a9b51d-2879-4523-9540-6ed033d1b5bc | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post201
This is a blog post reporting some preliminary work from the Anthropic Alignment Science team, which might be of interest to researchers working actively in this space. We'd ask you to treat these results like those of a colleague sharing some thoughts or preliminary experiments at a lab meeting, rather th... |
9571eb9c-ac82-4884-80b7-9162091f3c92 | trentmkelly/LessWrong-43k | LessWrong | Link: Facing the Mind-Killer
I've long opposed discussing politics on Less Wrong. Elsewhere, however, I have been known to gaze into the abyss; and so it came to be that I wrote a handful of blog posts of the Oxford Libertarian Society Blog. I had the deliberate intention of bring a little bit of rationality into poli... |
a260ee57-4dac-4840-8f6f-7d9dba40d058 | trentmkelly/LessWrong-43k | LessWrong | Can Social Dynamics Explain Conjunction Fallacy Experimental Results?
Is there any conjunction fallacy research which addresses the alternative hypothesis that the observed results are mainly due to social dynamics?
Most people spend most of their time thinking in terms of gaining or losing social status, not in term... |
4dfe32bb-3930-4a6d-9f7a-9befb1dfa3d2 | trentmkelly/LessWrong-43k | LessWrong | The Territory
I don’t know if you’ve ever had the experience of hiking in the wilderness with a map and compass but no cell service. I recommend it, if you haven’t.
I somehow did not quite all-the-way understand what a map even is until I was lost on my own under these circumstances. I knew in a “factual knowledge” ... |
1be4624d-9b24-4a8a-b6a5-cb09646c9445 | trentmkelly/LessWrong-43k | LessWrong | Recommended Rationalist Resources
I thought Recommended Rationalist Reading was very useful and interesting. Now we have voting and threading it seems a good time to comprehensively gather opinions on online material.
Please suggest high-quality links related to or useful for improving rationality. It could be a blog... |
7ea461b8-df98-4692-80c9-df1b7a2bb8f5 | trentmkelly/LessWrong-43k | LessWrong | "Giftedness" and Genius,
Crucial Differences
This essay by Arthur Jensen is from an old book .
The genius has limits. A simple answer, and undoubtedly true. But my
assignment here is to reflect on the much more complex difference
between intellectual giftedness and genius, using the latter term in its
original sense,... |
a119179f-ab2a-400c-ab3e-023138efa8d4 | trentmkelly/LessWrong-43k | LessWrong | Retrieval Augmented Genesis II — Holy Texts Semantics Analysis
Intro
In my last post on Lesswrong I shared ragenesis.com platform, along with a general description of the thought process behind the platform's development. Fundamentally, RAGenesis is a search and research AI driven software tool dedicated to 5 of the m... |
dc728562-3029-4239-8da8-6f96abd60ca0 | trentmkelly/LessWrong-43k | LessWrong | Southern California Meetup May 21, Weekly Irvine Meetups on Wednesdays
Southern California is about to have a lot more meetups. To get the ramp up started, there will be a meet up at this IHOP across the street from John Wayne Airport in Irvine, on Saturday May 21 from 2:00 to 8:00. We will likely be at the upstairs t... |
d3e0b437-ce95-4239-8eef-75b94101cbf5 | trentmkelly/LessWrong-43k | LessWrong | Careful with Caching
A friend recently shared this graphic on Facebook:
(I've added the "wrong" overlay.)
This is clearly incorrect: there's no way Massachusetts has more prisons than colleges. (MA actually has the largest ratio of colleges to prisons in the US.) After putting a link to the original source in the ... |
71900167-6574-44c2-a014-f595ee89dc08 | trentmkelly/LessWrong-43k | LessWrong | Y2K: Successful Practice for AI Alignment
This is a linkpost for Martyn Thomas's What Really Happened in Y2K and its associated talk recording.
Some spoke of a terrible catastrophe coming inevitably.
They said the death toll could be in the millions, if not higher.
An alarm was raised, but few paid attention.
But ... |
bfe6a8ab-465a-447b-bee7-b338db70feda | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Refine's Second Blog Post Day
Yesterday was the second blog post day at Refine. It came after the first week of research and iteration on incubatees ideas and questions.
It still went well. There are less posts than the first day, but even the incubatees that didn't write a post ended up working on interesting resear... |
f5af7758-a49a-4c0c-b9bc-00c1d7d0b932 | trentmkelly/LessWrong-43k | LessWrong | [SEQ RERUN] How Much Evidence Does It Take?
Today's post, How Much Evidence Does It Take? was originally published on 24 September 2007. A summary (taken from the LW wiki):
> If you are considering one hypothesis out of many, or that hypothesis is more implausible than others, or you wish to know with greater conf... |
d2589c46-7906-4481-9234-9a38d0f8fb83 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | [AN #108]: Why we should scrutinize arguments for AI risk
Alignment Newsletter is a weekly publication with recent content relevant to AI alignment around the world. Find all Alignment Newsletter **[resources here](http://rohinshah.com/alignment-newsletter/)**. In particular, you can look through **[this spreadsheet](... |
55dbc117-5322-4335-a2f7-24177a5268cf | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Goals of model vs. goals of simulacra?
In classical examples of misaligned AIs (like the famous paperclip maximiser ones), we tend to view the gradient descent training as the step in which an AI gets imbued with some overarching goal that it might then pursue to its bitter conclusion. However, the currently dominatin... |
cbb8120f-63ad-41b4-b2ad-05564885b104 | trentmkelly/LessWrong-43k | LessWrong | Ng and LeCun on the 6-Month Pause (Transcript)
This is a transcript a brief chat between Andrew Ng and Yann LeCun on the prospect of the 6-Month Pause. They are both highly skeptical of the pause and more generally of "AI doomerism". Yann also briefly touches on why he thinks LLMs won't scale to AGI.
Full video can... |
bda4139f-2554-4b5c-a73f-d79a9c14b337 | trentmkelly/LessWrong-43k | LessWrong | Training Regime Day 6: Seeking Sense
Introduction
Seeking sense is very easy to describe and very hard to actually do in practice. The basis of this technique is a claim is that the things that go on in other people's heads make sense from their perspective and that it's valuable to try to figure out what's happening ... |
05ac2423-e73a-4b74-8f0e-1c2f78885056 | trentmkelly/LessWrong-43k | LessWrong | Living with Rats in College
When I was in college, I rented a group house with 5 other nerds. There were 5 bedrooms to divide among the 6 of us, so I negotiated an agreement where I paid less rent in exchange for sleeping in the hallway. This wasn't as bad as it sounds. My bedroom wasn't hallway-sized. It was bedroom-... |
1f313fc4-7a95-477e-977e-916447cf4d29 | trentmkelly/LessWrong-43k | LessWrong | Why has no one compared Covid-19 and Vaccine Risks?
Currently, hundreds of millions of people face the choice to
(1) get vaccinated against Covid-19 (maybe choosing between different vaccines)
or
(2) do nothing
Given the fact that both the vaccines and contracting Covid-19 pose risks (in a sense that they may ... |
9db9586d-c4a9-4ff6-8122-62b867f8d3de | trentmkelly/LessWrong-43k | LessWrong | A sufficiently paranoid paperclip maximizer
A hypothetical scenario of a non-friendly but sufficiently paranoid AGI saving humanity.
----------------------------------------
The devil was laughing as the last of mankind showed up in his dominion. Then he stopped, because after them came some kind of machines that be... |
9cb86886-9e5e-4789-a949-45f0a12edfbd | trentmkelly/LessWrong-43k | LessWrong | Map of the AI Safety Community
I have made a map of the AI Safety Community!
The map is greatly inspired by the map of the rationalist community made by Scott Alexander.
There are bound to be omissions and misunderstandings, and I will be grateful for any corrections. I promise that I will incorporate the feedback i... |
8c479996-8b78-4839-b55a-743d35ea2fc6 | trentmkelly/LessWrong-43k | LessWrong | AI #102: Made in America
I remember that week I used r1 a lot, and everyone was obsessed with DeepSeek.
They earned it. DeepSeek cooked, r1 is an excellent model. Seeing the Chain of Thought was revolutionary. We all learned a lot.
It’s still #1 in the app store, there are still hysterical misinformed NYT op-eds and... |
2b676b50-0782-4eab-a0fe-89dcd79a63ad | trentmkelly/LessWrong-43k | LessWrong | I need to understand more about...
... the general take on climate change here.
Please read a little more before voting this down - I am not looking to initiate a debate on climate change - merely to understand what goes on when it is mentioned.
Disclosure: I am personally concerned about the impact of climate chang... |
886cb8bd-774a-400b-91c1-f915189f7c23 | trentmkelly/LessWrong-43k | LessWrong | Notifications Received in 30 Minutes of Class
Introduction
If you are choosing to read this post, you've probably seen the image below depicting all the notifications students received on their phones during one class period. You probably saw it as a retweet of this tweet, or in one of Zvi’s posts. Did you find this d... |
9cad0af5-9845-4839-818a-b5ea8720ebbb | trentmkelly/LessWrong-43k | LessWrong | Menstrual cycle effects—Clue study summary and commentary
Two months ago, I posted a question asking how LessWrong users optimize their productivity with respect to their menstrual cycle. User AM helpfully pointed me to a 2021 Nature Human Behavior study Daily, weekly, seasonal, and menstrual cycles in women's mood, b... |
6fc21455-c8b6-41f5-9b8e-545cc2141943 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Getting GPT-3 to predict Metaculus questions
Can GPT-3 predict real world events? To answer this question I had GPT-3 predict the likelihood for every binary question ever resolved on Metaculus.
Predicting whether an event is likely or unlikely to occur, often boils down to using common sense. It doesn't take a geniu... |
187e5cb4-0964-42d2-9a87-115d4e0da462 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Denver Area LW February Meetup
Discussion article for the meetup : Denver Area LW February Meetup
WHEN: 07 February 2017 07:00:00PM (-0700)
WHERE: 4955 South Ulster Street #103, Denver, CO
Darcy's Pub. We should be in the back room.
Join us to socialize and discuss Rationalist topics! We usually have semi... |
47bce2d6-e871-489a-8192-6ba17af05675 | trentmkelly/LessWrong-43k | LessWrong | Strategic Reliablism
Epistemology and the Psychology of Human Judgment by Michael Bishop and J.D. Trout
----------------------------------------
Ch 1. Laying Our Cards on the Table
Epistemology as a discipline needs to start offering practical advice for living. Defective epistemologies can compromise one's abil... |
f572165a-be79-4037-82d7-911e9d4b022d | trentmkelly/LessWrong-43k | LessWrong | Weekly LW Meetups
This summary was posted to LW Main on February 17th. The following week's summary is here.
The following meetups take place in cities with regular scheduling, but involve a change in time or location, special meeting content, or simply a helpful reminder about the meetup:
* Amsterdam - Meetup #12 ... |
2d53c569-f56f-4e64-8585-c9a32ac2a7ad | trentmkelly/LessWrong-43k | LessWrong | Grading my 2020 predictions
[Edit: when I copy-pasted from my blog, the colors didn't show up, and I'm not sure how to make them show up; sorry!]
In December 2019, I made 132 probabilistic predictions for 2020. As promised, I’ve come back to evaluate them on three criteria: calibration, personal optimism/pessimism, a... |
170bfe26-1399-48f9-bad8-569f0c332eb5 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Interviews with 97 AI Researchers: Quantitative Analysis
TLDR: Last year, Vael Gates interviewed 97 AI researchers about their perceptions on the future of AI, focusing on risks from advanced AI. Among other questions, researchers were asked about the alignment problem, the problem of instrumental incentives, and thei... |
fbdae8c1-baf5-420e-b06f-789de013a0bf | trentmkelly/LessWrong-43k | LessWrong | Checklists
Checklists are a rationality technique, mentioned previously on OB. Everyone knows this, but we don't hear about them as often as we should, possibly because they seem prosaic and boring.
In the context of doing something over and over, there is a checklist improvement cycle.
* You try to make the thing ... |
4bc4c915-ce77-4cdf-ad6d-8c522a2f3b3a | trentmkelly/LessWrong-43k | LessWrong | Friendliness in Natural Intelligences
The challenge of friendliness in Artificial Intelligence is to ensure how a general intelligence will be of utility instead of being destructive or pathologically indifferent to the values of existing individuals or aims and goals of their creation. The current provision of comput... |
a8c2a228-83b3-4e44-828e-0eb6685b203f | trentmkelly/LessWrong-43k | LessWrong | Why the Problem of the Criterion Matters
I think the problem of the criterion is really, really important. So important that I'm willing to risk annoying people by bringing it up all the time. I think it matters because a failure to grok it is at the heart of why we fail to notice we're confused and struggle to become... |
39974f2d-df2f-4aae-bed5-f294139019e2 | trentmkelly/LessWrong-43k | LessWrong | Good luck, Mr. Rationalist
Is there any rationalist equivalent of "good luck"? I've tried a few variants, such as "work well", "knock them dead", "we're with you" and certain situation-specific phrasings, but haven't found anything that worked generally - though a hearty "may all the gods of Olympus be with you!" can ... |
bdaca0c1-1440-439b-8ac1-fa00dd1573bf | trentmkelly/LessWrong-43k | LessWrong | Detecting Strategic Deception Using Linear Probes
Can you tell when an LLM is lying from the activations? Are simple methods good enough? We recently published a paper investigating if linear probes detect when Llama is deceptive.
Abstract:
> AI models might use deceptive strategies as part of scheming or misaligned... |
b43874d1-f68f-466a-99b4-5865578ccca0 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | A Defense of Work on Mathematical AI Safety
AI Safety was, a decade ago, nearly synonymous with obscure mathematical investigations of hypothetical agentic systems. Fortunately or unfortunately, this has largely been overtaken by events; the successes of machine learning and the promise, or threat, of large language m... |
8c87902f-38d7-4453-b3e5-11a162da6131 | StampyAI/alignment-research-dataset/blogs | Blogs | Multipolar research questions
*By Katja Grace, 11 February 2015*
The [Multipolar AI workshop](http://aiimpacts.org/event-multipolar-ai-workshop-with-robin-hanson/ "Event: Multipolar AI workshop with Robin Hanson") we ran a fortnight ago went well, and we just put up a [list of research projects](http://aiimpacts.org... |
8be56bf3-ff4c-406f-829c-c97614fd15f9 | trentmkelly/LessWrong-43k | LessWrong | Love is Love, Science is Fake
Previously in the sequence: Coupling for Decouplers — Intro
Cross-posted from SecondPerson.dating.
----------------------------------------
In my essay on "dating advice", we discussed the dating books full of recipe-like instructions for your romantic life. These recipes don’t really ... |
90d6ee71-7971-4b27-a2e3-611a2c59f947 | trentmkelly/LessWrong-43k | LessWrong | GPT-4 and ASCII Images?
Does anyone know whether GPT-4 successfully generates ASCII art?
GPT-3.5 couldn't:
the hell
Which makes sense, cause of the whole words-can't-convey-phenomena thing.
I'd expect multimodality to solve this problem, though? |
1d11dd8f-ed91-462e-893a-a3ed30067187 | trentmkelly/LessWrong-43k | LessWrong | Hilbert's Triumph, Church and Turing's failure, and what it means (Post #2)
We've specified what the definition of the Church-Turing Thesis means, though we've really specified 2 Church-Turing Theses, and the purist one is addressed in the 2.5 case, while a modified version handles the 2.6 case. We also specified what... |
81f4b14d-c91b-45b6-8400-a914feca6a14 | trentmkelly/LessWrong-43k | LessWrong | [SEQ RERUN] The Ultimate Source
Today's post, The Ultimate Source was originally published on 15 June 2008. A summary (taken from the LW wiki):
> There is a school of thought in philosophy that says that even if you make a decision, that still isn't enough to conclude that you have free will. You have to have been... |
4f68e4ba-4c79-4592-8426-71ebb85e6f8d | StampyAI/alignment-research-dataset/blogs | Blogs | ML Safety Newsletter #1
Welcome to the 1st issue of the ML Safety Newsletter. In this edition, we cover:
* various safety papers submitted to ICLR
* results showing that discrete representations can improve robustness
* a benchmark which shows larger models are more likely to repeat misinformation
* a benchmark for d... |
d570ec74-8fa8-43ec-95fe-8819c63c2e74 | trentmkelly/LessWrong-43k | LessWrong | I want to save myself
Related to: People who want to save the world
I have recently been diagnosed with cancer, for which I am currently being treated with good prognosis. I've been reevaluating my life plans and priorities in response. To be clear, I estimate that the cancer is responsible for much less than half t... |
22adc5bd-723d-4ef5-826d-d5623e219d9b | trentmkelly/LessWrong-43k | LessWrong | "... than average" is (almost) meaningless
Recently, I was talking to a friend who hadn't seen me in a while. They mentioned that my hair had grown noticeably, and then asked whether my hair grew fast or slow. I said that my hair growth was probably around average, but upon consideration, I realized that statement was... |
05459819-f4e2-4e9b-8d16-54e0b12f92d4 | trentmkelly/LessWrong-43k | LessWrong | Meta-error: I like therefore I am
I like Scott’s post on what LessWrong has learned in its lifetime. In general I approve of looking back at your past misunderstandings and errors, and trying to figure out what you did wrong. This is often very hard, because it’s hard to remember or imagine what absurd thoughts (or ab... |
265b191b-43e0-4cc0-bbc1-710e461ac65e | trentmkelly/LessWrong-43k | LessWrong | Convoy Continued
Previously: Convoy, also coverage from weekly post.
We continue to have ourselves a convoy, as well as attempts at similar additional convoys elsewhere.
We also have some rather startling actions on the part of the Canadian government, which made the ongoing situation impossible to ignore or disenta... |
a08b8ad2-39cf-4efa-a7c6-d0b5dcbe25e2 | trentmkelly/LessWrong-43k | LessWrong | Make Putanumonit Great (for the first time)
Good news: 2016 is coming to a close.
Bad news: “2016” is an arbitrary construct without causal agency. The decline of civilization is not the calendar’s fault, and it’s unlikely to get any better in 2017. Only one thing is guaranteed to get better in 2017 – this blog. Beca... |
6398880f-04e0-4702-9127-8bb3f129b29c | trentmkelly/LessWrong-43k | LessWrong | Why do you have a personal rationalist blog instead of using LessWrong? Why don't you cross-post all your relevant blog posts?
to those it applies
also see: Site Guide: Personal Blogposts vs Frontpage Posts |
00858f57-d117-45a3-8de3-53612a581f7e | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Eli Lifland on Navigating the AI Alignment Landscape
Recently I had a conversation with [Eli Lifland](https://forum.effectivealtruism.org/users/elifland) about the AI Alignment landscape. Eli Lifland has been a forecaster at [Samotsvety](https://samotsvety.org/) and has been investigating said landscape.
I’ve known E... |
16f9f8fc-b17a-4432-b510-0dfacdfd5549 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Alignment Frame/Exercise: Building The Puzzle of Alignment
(5-6 min read time without exercise instructions)
***Motivation for post**: Finding a way to map mathematics & areas of academic research to the alignment problem seems to be a core part of theoretical alignment and agent foundations work. Knowing how to fram... |
20453a09-e28d-465c-a2b2-98e7b1aa8315 | trentmkelly/LessWrong-43k | LessWrong | On Intentionality, or: Towards a More Inclusive Concept of Lying
Intentional actions may be contrasted with unintentional actions, which are generally the result of either instinct or accident. More precisely, we may say that unintentional actions are either the result of some unconscious cognitive process that culmin... |
7af727da-8667-4b02-bdb4-05f61d357ba7 | trentmkelly/LessWrong-43k | LessWrong | Superintelligence Strategy: A Pragmatic Path to… Doom?
I've been reading through this Superintelligence Strategy paper by Dan Hendrycks, Eric Schmidt, and Alexandr Wang. And, to me, it sounds like the authors are calling the current regime (MAIM aka a "Hot War") the "default" (which it probably is, tbh). But, also cal... |
d59f9fa8-d47a-4eca-bb72-6052e0015634 | trentmkelly/LessWrong-43k | LessWrong | How likely is it that US states or cities will prevent travel across their borders?
In contrast to quarantining people who are known to be infected, I want to know how likely it is that, (for instance) California doesn't allow people to enter and/or to leave, across her state borders.
Is there any precedent in US his... |
aaaf05d2-f2e9-4bab-b0a9-8c798788ff18 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | We're Redwood Research, we do applied alignment research, AMA
Redwood Research is a longtermist organization working on AI alignment based in Berkeley, California. We're going to do an AMA this week; we'll answer questions mostly on Wednesday and Thursday this week (6th and 7th of October). Buck Shlegeris, Bill Zito, ... |
78b02769-e0dd-4f09-9ea6-2b3f95b5f569 | trentmkelly/LessWrong-43k | LessWrong | Chapter 74: SA, Escalation of Conflicts, Pt 9
Harry walked forward a step, then another step, until a sense of unease began to pervade him, a disquiet in his nerves.
He said nothing, lifted no hand; the pervading sense of unease would say it for him.
From behind the closed door of the office came a whisper, carrying... |
72234e29-714c-4eda-9122-b600f033dbf4 | trentmkelly/LessWrong-43k | LessWrong | Emotional issues often have an immediate payoff
1. It can be extremely valuable to view emotional issues as having an immediate payoff. For example, depression, anxiety, insecurity, failure at work, failure in romance, muscle tension, chronic pain, etc. can help one avoid fears or achieve unconscious goals.
2. But v... |
acfc4544-d994-418c-8fd5-0c4f793c0704 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | New tool for exploring EA Forum, LessWrong and Alignment Forum - Tree of Tags
Explore the hidden treasures of the forums
==========================================
With this tool you can zoom in on your f... |
b795b184-72f5-4e87-a353-e7d0670b9252 | StampyAI/alignment-research-dataset/arbital | Arbital | Ignorance prior
An ignorance prior is a a [prior](https://arbital.com/p/1rm) [probability function](https://arbital.com/p/1zj) on some problem of interest, usually with the intended properties of being simple to describe and facilitating good learning from the evidence. A classic example would be the [inductive prior... |
5c6ceb2b-dd21-42dc-a756-2171eacbd091 | trentmkelly/LessWrong-43k | LessWrong | If Clarity Seems Like Death to Them
> "—but if one hundred thousand [normies] can turn up, to show their support for the [rationalist] community, why can't you?"
>
> I said wearily, "Because every time I hear the word community, I know I'm being manipulated. If there is such a thing as the [rationalist] community, I'... |
019cba27-a0d3-4efa-80c2-8fe4b381e202 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Is there any literature on using socialization for AI alignment?
Hello,
I was recently thinking about the question of how humans achieve alignment with each other over the course of our lifetime, and how that process could be applied to an AGI.
For example, why doesn't everyone shop lift from the grocery store? A ... |
34fdd949-969d-456d-abec-7b67ee926506 | trentmkelly/LessWrong-43k | LessWrong | The New Riddle of Induction: Neutral and Relative Perspectives on Color
Nelson Goodman's "New Riddle of Induction" has previously been discussed on Less Wrong at http://lesswrong.com/lw/8fq/bayes and http://lesswrong.com/lw/mbr/grue . Briefly, this riddle shows that any attempt to make future predictions by generalizi... |
8f96407e-ea04-4d63-b0b7-85a5630d83a9 | trentmkelly/LessWrong-43k | LessWrong | Outline of Metarationality, or much less than you wanted to know about postrationality
There was a recent discussion on Facebook that led to an ask for a description of postrationality that isn't framed in terms of how it's different from rationality (or rather perhaps more a challenge that such a thing could not be p... |
32148c8e-ce38-4d45-b0b4-e95e14ccc040 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Rohin Shah: What’s been happening in AI alignment?
---
*While we haven’t yet built aligned AI, the field of alignment has steadily gained ground in the past few years, producing many useful outputs. In this talk, Rohin Shah, a sixth-year PhD student at UC Berkeley’s* [*Center for Human-Compatible AI (CHAI)*](https://... |
f8f4a101-ac03-4569-b9c3-b4940ca9644b | trentmkelly/LessWrong-43k | LessWrong | If someone you loved was experiencing unremitting suffering (related to a constellation of multi-dimensional factors and processes, those of which include anomalous states of consciousness and an iatrogenic mental health system), what would you think and what would you do?
Before I get into this more tangibly, I want ... |
368ee749-fe65-41af-ae81-9c05beb066f2 | trentmkelly/LessWrong-43k | LessWrong | The Nature of Offense
Recently, an extended discussion has taken place over the fact that a portion of comments here were found to be offensive by some members of this community, while others denied their offensive nature or professed to be puzzled by why they are considered offensive. Several possible explanations fo... |
bb7ac402-b4dc-47c9-a443-a54570877ef3 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | A Certain Formalization of Corrigibility Is VNM-Incoherent
Edit, 5/16/23: I think this post is beautiful, correct in its narrow technical claims, and practically irrelevant to alignment. This post presents an unrealistic picture of the role of reward functions in reinforcement learning, conflating "utility" with "rewa... |
89694476-9a1f-4912-b93f-d8fdf31066af | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "A transcribed excerpt I found interesting from the decision theory outtake from 80,000 Hours' second interview of Paul Christiano (starting at 14:00):Robert Wiblin: It seems like philosophers have not been terribly interested in these heterodox decision theories. They just seem to not have been persuaded, or not even ... |
89e7f506-4939-410a-85fe-da5ccb827465 | trentmkelly/LessWrong-43k | LessWrong | Four usages of "loss" in AI
Summary: What does it mean for a loss function to be "aligned with" human goals? I perceive four different concepts which involve "loss function" in importantly different ways:
1. Physical-loss: The physical implementation of a loss function and the loss computations,
2. Mathematical-lo... |
0ffd453f-d90c-4b4f-9fed-5ee68684aa51 | trentmkelly/LessWrong-43k | LessWrong | The Debtor's Revolt
Ben Hoffman describes the transition from a old world order to a debtor-aristocracy world order.
Consider a standard microeconomics model, where profits are good and debts represent some loss of that profit in interest; individual actions are taken based on whether they will be return more than th... |
185620a9-986a-4e79-887f-56d55cb7268f | StampyAI/alignment-research-dataset/arxiv | Arxiv | Identifying and Correcting Label Bias in Machine Learning
1 Introduction
---------------
Machine learning has become widely adopted in a variety of real-world applications that significantly affect people’s lives (Guimaraes and Tofighi, [2018](#bib.bib25); Guegan and Hassani, [2018](#bib.bib24)). Fairness in these ... |
90a2f677-4601-4845-a765-43a4d34b78c4 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Humans Who Are Not Concentrating Are Not General Intelligences
Recently, OpenAI came out with a [new language model](https://blog.openai.com/better-language-models/) that automatically synthesizes text, called GPT-2.
It’s disturbingly good. You can see some examples (cherry-picked, by their own admission) in OpenAI... |
a7397ce9-d617-4e35-9cf4-d8a15ab69f64 | trentmkelly/LessWrong-43k | LessWrong | AISafety.info: What are Inductive Biases?
AISafety.info writes AI safety intro content. We'd appreciate any feedback.
The inductive bias of a learning process is its tendency to learn particular kinds of patterns from data. It would help with AI alignment to know which learning processes have inductive biases towa... |
5444185e-2eea-4f54-a864-bbf4f4fbc04e | trentmkelly/LessWrong-43k | LessWrong | The Difficulties of Potential People and Decision Making
In connection to existential risk and the utility of bringing future people into being as compared with the utility of protecting those currently alive, I’ve been looking into the issues and paradoxes present in the ethics of potential persons. This has led to a... |
d47620e0-23c5-40a9-a133-fe1c35e72065 | trentmkelly/LessWrong-43k | LessWrong | [LINK] '3 Secrets of Wise Decision Making'
Personal Decision Making (textbook about applied decision-making, for the class I'm taking right now)
The reason the book struck me as interesting was the author is employed at the university I'm going to, so if I get stuck its possible I could go and talk to him myself, and... |
990c0313-acb3-45d6-82ca-c8fd5d8291c6 | trentmkelly/LessWrong-43k | LessWrong | January 2016 Media Thread
This is the monthly thread for posting media of various types that you've found that you enjoy. Post what you're reading, listening to, watching, and your opinion of it. Post recommendations to blogs. Post whatever media you feel like discussing! To see previous recommendations, check out the... |
7e91ab80-a25c-49b2-b7f8-26df332810bb | StampyAI/alignment-research-dataset/lesswrong | LessWrong | VNM expected utility theory: uses, abuses, and interpretation
*When interpreted convservatively, the von Neumann-Morgenstern [rationality axioms and utility theorem](http://en.wikipedia.org/wiki/Von_Neumann%E2%80%93Morgenstern_utility_theorem) are an indispensible tool for the normative study of rationality, deserving... |
136f3ca0-8a4f-451c-a17e-5121ff7f8b04 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | AISN #23: New OpenAI Models, News from Anthropic, and Representation Engineering
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 [here](https://newsletter.safe.ai/subscribe?utm_medium=web&... |
83d02242-6b12-454c-9dd8-9dd4ca4d5a6e | trentmkelly/LessWrong-43k | LessWrong | 2 of 10, not 3 total
There is no rule against commenting more than 3 times in a thread. Sorry if anyone has gotten this impression.
However, among the 10 "Recent Comments" visible in the sidebar at right, usually no more than 2, rarely 3, and never 4, should be yours. This is meant to ensure no one person dominates... |
4345caa8-095f-45c3-bc4e-da0353ff8910 | trentmkelly/LessWrong-43k | LessWrong | Why are we sure that AI will "want" something?
I have no doubt that AI will some day soon surpass humans in all aspects of reasoning, that is pretty obvious. It is also clear to me that will surpass humans in the ability to do something, should it "want" to do it. And if requested to do something drastic, it can accid... |
bfbad9cc-d5c5-45ac-93b7-e28f65f7f7b0 | trentmkelly/LessWrong-43k | LessWrong | Why Bayes? A Wise Ruling
Why is Bayes' Rule useful? Most explanations of Bayes explain the how of Bayes: they take a well-posed mathematical problem and convert given numbers to desired numbers. While Bayes is useful for calculating hard-to-estimate numbers from easy-to-estimate numbers, the quantitative use of Bayes ... |
f0e16b7d-81f7-466a-9416-c5a163060978 | trentmkelly/LessWrong-43k | LessWrong | Partial Transcript of the Hanson-Yudkowsky June 2011 Debate
So I'm currently trying to write an article on the nature of intelligence and the AI FOOM issue. I listened to the live debate between Eliezer Yudkowsky and Robin Hanson they did in June 2011, and thought taken together Eliezer's comments in that debate make ... |
40722cf6-5a8f-44ea-8516-d68d1721e1b6 | trentmkelly/LessWrong-43k | LessWrong | Rolling Thresholds for AGI Scaling Regulation
This is a plan for how ASI could be relatively safely developed.
Abstract: A plan that puts all frontier model companies on a unified schedule of model training, evaluation and approval, with regulatory compliance promoted through market access. This aims to combine (most... |
f61ed5ac-6b45-474f-8cea-d220d9880b8e | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | EfficientZero: human ALE sample-efficiency w/MuZero+self-supervised
["Mastering Atari Games with Limited Data"](https://arxiv.org/abs/2111.00210), Ye et al 2021:
>
> Reinforcement learning has achieved great success in many applications. However, sample efficiency remains a key challenge, with prominent methods re... |
6019f589-d23b-4aed-8bc1-c2f20ab14dfe | trentmkelly/LessWrong-43k | LessWrong | East Coast Megameetup II: Request for Talks
One thing that I'd like to see at the next megameetup is a few informal focused discussions on various topics related to the community, like meditation or diet.
People should prepare for them, but I don't think that say, powerpoints would be necessary. Just trying to encour... |
1edcd605-1983-41e6-88d4-f3dc520c4658 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | What Was Your Best / Most Successful DALL-E 2 Prompt?
Share successful ways the AI was able to generate your image in the most accurate way possible. This could mean certain key phrases you inserted or the way the prompt is constructed from start to finish. |
8afd254b-e599-4362-a3f2-b27626c4ce51 | trentmkelly/LessWrong-43k | LessWrong | Biosecurity and AI: Risks and Opportunities
Recent decades have seen massive amounts of biological and medical data becoming available in digital form. The computerization of lab equipment, digitization of medical records, and advent of cheap DNA sequencing all generate data, which is increasingly collected in large d... |
8f902782-b5ff-4036-b3a3-f4194b61784b | trentmkelly/LessWrong-43k | LessWrong | Greed Is the Root of This Evil
The SGD's greed, to be specific.
Consider a ML model being trained end-to-end from initialization to zero loss. Every individual update to its parameters is calculated to move it in the direction of maximal local improvement to its performance. It doesn't take the shortest path from whe... |
e02fa728-163e-4919-893d-10e0e44dd08a | trentmkelly/LessWrong-43k | LessWrong | Time turners, Energy Conservation and General Relativity
This post is a bit of entertainment for scientifically inclined Harry Potter fans.
Time turner from the Harry Potter series (and from the Eliezer Yudkowsky's venerable HPMoR fanfic) is a very useful device if you have some unfinished business in the recent past... |
aa880d47-defd-4f0e-8aa7-da977f0c2b45 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Sentience in Machines - How Do We Test for This Objectively?
Hello.
I'm new here, and this is my first post.
The central topic involves the notion of sentience in machines - "Can machines develop Sentience? What does Sentience mean in the context of machines? and How do you reliably test for it?"
Last year there wa... |
5feb7fb3-2858-4293-9959-36291031ac19 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Reading the ethicists 2: Hunting for AI alignment papers
Introduction
============
I'm back, reading more papers in ethics (but now also in philosophy and wherever the citations lead me). Unlike [last time](https://www.lesswrong.com/posts/D2biiCrF7E62kDbGA/reading-the-ethicists-a-review-of-articles-on-ai-in-the) when... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.