id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
b84d5456-cc2c-4a34-a0ca-f45058c1a7a1 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | The voyage of novelty
*[Metadata: crossposted from <https://tsvibt.blogspot.com/2023/01/the-voyage-of-novelty.html>. First completed January 17, 2023. This essay is more like research notes than exposition, so context may be missing, the use of terms may change across essays, and the text might be revised later; only ... |
6081c55d-44ef-4674-92fd-d9a4a61866e7 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | RAISE is launching their MVP
After over a year of [building the organisation](https://www.lesswrong.com/posts/ATQ23FREp9S4hpiHc/raising-funds-to-establish-a-new-ai-safety-charity) and working and reworking the product, I’m glad to announce that [RAISE](https://aisafety.info/) will begin publishing lessons on a regular... |
542a06c2-36f8-42e4-ac6c-5ea1f9b162f9 | trentmkelly/LessWrong-43k | LessWrong | The Human Alignment Problem for AIs
> If there was a truly confirmed sentient AI, nothing it said could ever convince me, because AI cannot be sentient.
Nothing to See Here
I suspect at least some will be nodding in agreement with the above sentiment, before realizing the intentional circular absurdity. There is ... |
e8ea35a1-0953-441c-8467-8adeb44fea87 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | My current workflow to study the internal mechanisms of LLM
This is a post to keep track my research workflow of studying LLM. Since I am doing it on my spare time, I want to keep my pipeline as simple as possible.
Step 1: Formulate a question for investigating model's behavior .
Step 2: Find the influential layer... |
7ba0cd80-fded-4cc5-b896-71e7375e2de7 | trentmkelly/LessWrong-43k | LessWrong | Delegative Inverse Reinforcement Learning
We introduce a reinforcement-like learning setting we call Delegative Inverse Reinforcement Learning (DIRL). In DIRL, the agent can, at any point of time, delegate the choice of action to an "advisor". The agent knows neither the environment nor the reward function, whereas th... |
cbe0a66b-e441-419d-b80e-7b21c0573737 | trentmkelly/LessWrong-43k | LessWrong | Request for rough draft review: Navigating Identityspace
Ignore this. After getting some much needed sleep I decided that the first half of this post, though necessary if I am to write the second half, isn't justified in being so complicated unless I actually use the added complications in solving the problems posed. ... |
3346c8f2-174d-448d-afc3-dd304b9cae99 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "Once upon a time I tried to tell my mother about the problem of expert calibration, saying: “So when an expert says they’re 99% confident, it only happens about 70% of the time.” Then there was a pause as, suddenly, I realized I was talking to my mother, and I hastily added: “Of course, you’ve got to make sure to appl... |
5681286e-febf-455a-b4fe-9ecd8286edc7 | trentmkelly/LessWrong-43k | LessWrong | Mistakes #6: Identifying reality slowly
(Mistakes #1, #2, #3, #4, #5)
One thing I was confused about when I was about eight was which things were ‘real’. It was not that I had any trouble distinguishing between the mundane objects around me—including my parents and a perplexed child psychologist—and my so-called ‘ima... |
41e07ae8-0881-4113-9dab-b945226d3ba6 | trentmkelly/LessWrong-43k | LessWrong | Has the effectiveness of fever screening declined?
On March 6th, jimrandomh predicted that the effectiveness of fever-screening for coronavirus would decline, because the virus would evolve to produce fever later or not at all. Has this happened in countries that have been liberally applying fever-screening? |
05a3f8df-596f-451d-a3a4-1f22854be0d2 | trentmkelly/LessWrong-43k | LessWrong | Effective hypocrisy?
You know what is cheap? Talk.
You know what is expensive? Action.
You know what is cost-effective? Hypocrisy.
At least if non-word actions are not much more effective than words, which seems right. Differences to the world you can make without communicating seem limited. And for communication... |
2a0d77dc-0d9f-4e70-9824-e0d780a8e790 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | What Makes an Idea Understandable? On Architecturally and Culturally Natural Ideas.
Midjourney generating a HD image of "a mediu... |
7e9653a7-e756-4f48-ad05-2e6d611a7149 | trentmkelly/LessWrong-43k | LessWrong | Near-term planning for secondary impacts of coronavirus lockdowns [assuming things don't blow up]
In this document, I attempt to discuss the impact of the coronavirus lockdown and how to prepare for it. This is not focused on the direct impact of coronavirus, but rather on the secondary impact of precautions that peop... |
f554b606-f978-4ef1-bf93-60237a784441 | trentmkelly/LessWrong-43k | LessWrong | Does Age Bring Wisdom?
I.
I turn 33 today. I can only hope that age brings wisdom.
We’ve been talking recently about the high-level frames and heuristics that organize other concepts. They’re hard to transmit, and you have to rediscover them on your own, sometimes with the help of lots of different explanations and ... |
6196100b-7003-4bc8-a071-0274c05c2a1c | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Finding Goals in the World Model
### *Produced As Part Of The SERI ML Alignment Theory Scholars Program 2022 Under* [*John Wentworth*](https://www.lesswrong.com/users/johnswentworth)
Introduction
============
This post works off the assumption that the first AGI comes relatively soon, and has an architecture which l... |
15d34912-1c9a-45c6-b01b-74bce0b90009 | trentmkelly/LessWrong-43k | LessWrong | The two insights of materialism
Preceded by: There just has to be something more, you know? Followed by: Physicalism: consciousness as the last sense.
Contents: 1. An epistemic difficulty 2. How and why to be a materialist
An epistemic difficulty
Like many readers of this blog, I am a materialist. Like many s... |
a9a9221d-e14a-4eff-820f-8f3981ca904a | trentmkelly/LessWrong-43k | LessWrong | Views on when AGI comes and on strategy to reduce existential risk
Summary: AGI isn't super likely to come super soon. People should be working on stuff that saves humanity in worlds where AGI comes in 20 or 50 years, in addition to stuff that saves humanity in worlds where AGI comes in the next 10 years.
Thanks to ... |
c5ec5564-8020-4c1f-ad90-edb61b9cb74e | trentmkelly/LessWrong-43k | LessWrong | Using Betting Markets to Make Decisions
Epistemic status: This seems sufficiently basic that it seems like someone would already have developed a general theory, but I haven’t managed to find anything. If you know of anything like this please let me know. This post copied over from here.
Say you want to make some rea... |
b4f99520-2f45-417c-8fad-d8a8442e045a | StampyAI/alignment-research-dataset/blogs | Blogs | Will MacAskill on normative uncertainty
 [Will MacAskill](http://www.williammacaskill.com/) recently completed his DPhil at Oxford University and, as of October 2014 will be a Research Fellow at Emmanuel College, Cambri... |
91c89bb5-101d-4b23-847b-1498f672f44c | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Two reasons we might be closer to solving alignment than it seems
I was at an AI safety retreat recently and there seemed to be two categories of researchers:
1. Those who thought most AI safety research was useless
2. Those who thought *all* AI safety research was useless
This is a darkly humurous anecdote illustra... |
0003f5c2-deac-4574-a4be-fbdfa5763e05 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | AXRP Episode 1 - Adversarial Policies with Adam Gleave
[Google Podcasts link](https://podcasts.google.com/feed/aHR0cHM6Ly9heHJwb2RjYXN0LmxpYnN5bi5jb20vcnNz/episode/ZjgxZWU3YjYtMTdmMS00NmQ1LWExZjctZGY0NGQzZGJlNjBi)
This podcast is called AXRP, pronounced axe-urp and short for the AI X-risk Research Podcast. Here, I (... |
a47fe9e8-64cb-4ba5-87a2-989f3a6fef81 | trentmkelly/LessWrong-43k | LessWrong | The Evolutionary-Cognitive Boundary
I tend to draw a very sharp line between anything that happens inside a brain and anything that happened in evolutionary history. There are good reasons for this! Anything originally computed in a brain can be expected to be recomputed, on the fly, in response to changing circumst... |
4551b598-d8cb-47f3-966b-b01dc85f61bd | trentmkelly/LessWrong-43k | LessWrong | Living in Many Worlds
Some commenters have recently expressed disturbance at the thought of constantly splitting into zillions of other people, as is the straightforward and unavoidable prediction of quantum mechanics.
Others have confessed themselves unclear as to the implications of many-worlds for planning: If you... |
e08a5d2c-14cc-4843-a689-00db9c541652 | trentmkelly/LessWrong-43k | LessWrong | Covid 8/6: The Case of the Missing Data
Last time: Covid 7/30: Whack a Mole
Testing. One could sum up the entire failure on Covid-19 as a failure of testing.
We have failed to test individuals. We have failed to ramp up testing capacity fast enough. We have failed to deliver test results within the time frame necess... |
b8b9bced-d170-4549-859f-a8dcbeba18e6 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | ISO: Name of Problem
I'm looking for a name for a problem. I expect it already has one, but I don't know what it is.
The problem: suppose we have an AI trying to learn what people want - e.g. an IRL variant. Intuitively speaking, we point at a bunch of humans and say “figure out what they want, then do that”. A few p... |
fe1a2996-9b15-4d2f-ad90-6df73c13a044 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | A possible check against motivated reasoning using elicit.org
**Are you worried you may be engaging in** [**motivated reasoning**](https://www.lesswrong.com/tag/motivated-reasoning)**, rationalization ... or committing other reasoning fallacies?**
**I propose** the following epistemic check using [Elicit.org](eli... |
8c9ade1c-3d32-4703-b8d9-2051beaca935 | trentmkelly/LessWrong-43k | LessWrong | Noragami
Noragami is a 12-episode anime that aired in 2014. The first 11 episodes are fine. I liked them but I didn't understand why until the 12ᵗʰ episode. The 12ᵗʰ episode blew me away.
Japan today has many layers of foreign influence. The most recent is the American occupation. Before that was the Meiji Restoratio... |
7b41ab2f-eead-4598-81d4-012859fe6a76 | trentmkelly/LessWrong-43k | LessWrong | o3 Will Use Its Tools For You
OpenAI has finally introduced us to the full o3 along with o4-mini.
> Greg Brockman (OpenAI): Just released o3 and o4-mini! These models feel incredibly smart. We’ve heard from top scientists that they produce useful novel ideas. Excited to see their positive impact on people’s daily liv... |
bdc2514d-33c5-48ad-afdc-47802939de30 | trentmkelly/LessWrong-43k | LessWrong | Time in Cartesian Frames
This is the twelfth and final post in the Cartesian Frames sequence. Read the first post here.
Up until now, we have (in the examples) mostly considered agents making a single choice, rather than acting repeatedly over time.
The actions, environments, and worlds we've considered might be ext... |
754a7197-5d97-4d7e-acbb-05a6b5781d52 | trentmkelly/LessWrong-43k | LessWrong | My prediction for Covid-19
In writing a reply to Zvi’s recent post on Covid-19 predictions I realized I had been procrastinating on writing out my thoughts on Covid-19 in full. In this post I will explain in detail how I think about the Covid-19 countermeasures for the next few months, and explain why I think all this... |
5e8bb07b-a2f9-4be5-85c2-21e20bb346ba | trentmkelly/LessWrong-43k | LessWrong | Crazy Ideas Thread, Aug. 2015
This thread is intended to provide a space for 'crazy' ideas. Ideas that spontaneously come to mind (and feel great), ideas you long wanted to tell but never found the place and time for and also for ideas you think should be obvious and simple - but nobody ever mentions them.
This threa... |
7cf9c041-1758-4c71-b147-3ffb1785c0cc | trentmkelly/LessWrong-43k | LessWrong | Help: Info on intelligence-focused genetic engineering?
This is a request for analyses and info on intelligence-related genetic engineering; how far away is genetic engineering designed to increase intelligence? What's the scientific/cultural/technological landscape? Are the obstacles funding-related or technological?... |
74e22330-483c-49dd-b8bf-c4ea9232737c | trentmkelly/LessWrong-43k | LessWrong | 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 ... |
08dcf95b-7578-4ca1-bb3d-65cd70ece407 | trentmkelly/LessWrong-43k | LessWrong | Rationality Quotes July 2012
Here's the new thread for posting quotes, with the usual rules:
* Please post all quotes separately, so that they can be voted up/down separately. (If they are strongly related, reply to your own comments. If strongly ordered, then go ahead and post them together.)
* Do not quote your... |
1e227676-2cd2-4e0e-9170-237c7fe18be9 | trentmkelly/LessWrong-43k | LessWrong | Terminal Values and Instrumental Values
On a purely instinctive level, any human planner behaves as if they distinguish between means and ends. Want chocolate? There's chocolate at the Publix supermarket. You can get to the supermarket if you drive one mile south on Washington Ave. You can drive if you get into th... |
82790c43-36d8-47f5-8083-293be2aca4cb | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | AI, Robot - DeepMind: The Podcast (S1, Ep4)
[Music]
the idea of creating artificial
creatures has obsessed us for millennia
but for the purposes of this exercise
before we go any further I'm going to
ask you to purge any of the following
thoughts from your head checklist
Hephaestus expelled from Olympus and
then built... |
97cf3dee-49da-4717-a23c-96bcf348dac9 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | What “defense layers” should governments, AI labs, and businesses use to prevent catastrophic AI failures?
Rethink Priorities is working on a project called ‘Defense in Depth Against Catastrophic AI Failures’. “[Defense in depth](https://forum.effectivealtruism.org/tag/defense-in-depth)” refers to the use of multiple... |
237b9511-fc13-438a-ae5f-df13696b5ff8 | trentmkelly/LessWrong-43k | LessWrong | Towards no-math, graphical instructions for prediction markets
You are probably familiar with prediction markets. As of the last general update, Robin Hanson notes that while there is increasing research interest and a few general platforms are coming soon, the real goal of implementing inside organizations remains el... |
a19660f1-5b0b-4ba9-8bbb-2e3f4ccbecb1 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Why I’m not working on {debate, RRM, ELK, natural abstractions}
[For background & spelling out the acronyms in the title, see: [Debate (AI safety technique)](https://www.lesswrong.com/tag/debate-ai-safety-technique-1), [Recursive Reward Modeling](https://deepmindsafetyresearch.medium.com/scalable-agent-alignment-via-r... |
334256eb-5e15-43aa-b0b7-703455154383 | trentmkelly/LessWrong-43k | LessWrong | New LW Meetup: Kyiv
This summary was posted to LW Main on December 23rd. The following week's summary is here.
New meetups (or meetups with a hiatus of more than a year) are happening in:
* Kyiv: 29 December 2016 18:15 (+0200): 29 December 2016 06:15PM
The following meetups take place in cities with regular schedu... |
dbc1ee14-3dec-4ef3-a4e8-b56a9a40bcb8 | trentmkelly/LessWrong-43k | LessWrong | Doing Important Research on Amazon's Mechanical Turk?
There seems to be many important questions that need research, from the mundane (say, which of four slogans for 80,000 Hours people like best) to the interesting (say, how to convince people to donate more than they otherwise would). Unfortunately, it's difficult ... |
b9894667-a539-4d46-89b2-3fe051b0c5de | trentmkelly/LessWrong-43k | LessWrong | (Virtual) Employment Open Thread
tl;dr: Some people on LW have a hard time finding worthwhile employment. Share advice and help them out!
Working sucks. I'd rather not work. But alas, a lot of the time, we have to choose between working and starvation. At the very least I'd like to minimize work. I'd like to work som... |
97797298-567a-4dc6-ac44-f6e7cf98216d | trentmkelly/LessWrong-43k | LessWrong | Andrew Ng dismisses UFAI concerns
http://fusion.net/story/54583/the-case-against-killer-robots-from-a-guy-actually-building-ai/
> “I don’t work on preventing AI from turning evil for the same reason that I don’t work on combating overpopulation on the planet Mars,” he said. “Hundreds of years from now when hopeful... |
760513b4-9cd4-42d3-a53a-a69640950006 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Blake Richards on Why he is Skeptical of Existential Risk from AI
*(*[*crossposted*](https://www.lesswrong.com/posts/dtpLLRbKjoDXs7b2a/blake-richards-on-why-he-is-skeptical-of-existential-risk) *from LW)*
I have recently interviewed Blake Richards, an Assistant Professor in the Montreal Neurological Institute and the... |
e485f677-c9df-4e8e-aae5-108e82087e30 | trentmkelly/LessWrong-43k | LessWrong | Open Thread: December 2011
If it's worth saying, but not worth its own post (even in Discussion), then it goes here.
If continuing the discussion becomes impractical, that means you win at open threads; a celebratory top-level post on the topic is traditional. |
985eacde-03f8-4b44-b29b-bb007dfdd702 | trentmkelly/LessWrong-43k | LessWrong | UDT agents as deontologists
One way (the usual way?) to think of an agent running Updateless Decision Theory is to imagine that the agent always cares about all possible worlds according to how probable those worlds seemed to the agent's builders when they wrote the agent's source code[see added footnote 1 below]. In... |
bf9dab21-c8ed-41dd-b66a-d7c2ae33b973 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Manifund x AI Worldviews
Manifund is hosting impact certificates for the Open Philanthropy AI Worldviews Contest!
* **Open Phil** is running this [essay contest on AI safety](https://www.openphilanthropy.org/open-philanthropy-ai-worldviews-contest/), with $225k in prizes
* **Essay authors** can now sell shares of pot... |
b445cad6-e359-4e99-b007-68e173268c4c | trentmkelly/LessWrong-43k | LessWrong | Outline: The Rectifying of Maps
Epistemic status: a hunch that having something like this matters a lot, and a basic outline for people to play with
In my culture, we take updating when we notice an error in our thinking very seriously. We have a cultural toolkit for it. In those moments, we want the comfort of a fam... |
8525d895-c209-43c7-916f-ee7bded9d945 | trentmkelly/LessWrong-43k | LessWrong | Will reason ever outrun faith?
Recently, a video produced by Christians claimed that the future world would be Muslim. It hit 10 million hits in YouTube. The alarming demographics presented were proven mostly false and exaggerated both by BBC and Snopes. Yet, religion is such a powerful self-replicating memeplex that ... |
9f41d009-86fb-4c7c-b8a3-58ae253bb3b6 | trentmkelly/LessWrong-43k | LessWrong | Consistent extrapolated beliefs about math?
My beliefs about the integers are a little fuzzy. I believe the things that ZFC can prove about the integers, but there seems to be more than that. In particular, I intuitively believe that "my beliefs about the integers are consistent, because the integers exist". That's an... |
76357afb-44df-4d9e-a24e-b9d63ba3761b | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post2255
Here’s a description of the project Redwood Research is working on at the moment. First I’ll say roughly what we’re doing, and then I’ll try to explain why I think this is a reasonable applied alignment project, and then I’ll talk a bit about the takeaways I’ve had from the project so far. There are a bun... |
60376dc0-bf52-421c-a976-08a4d40575e1 | trentmkelly/LessWrong-43k | LessWrong | List your AI X-Risk cruxes!
[I'm posting this as a very informal community request in lieu of a more detailed writeup, because if I wait to do this in a much more careful fashion then it probably won't happen at all. If someone else wants to do a more careful version that would be great!]
By crux here I mean some unc... |
d2f2bc4b-0114-4901-a6bd-31dae0217d69 | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | AiTech Agora - Jie Yang: ARCH - Know What Your Machine Doesn't Know
so it is better
afterwards there will be room for
questions in the middle so there will be
a designated uh spot
in the middle of the presentation for
questions okay thanks
okay uh the floor is yours okay
thank you akadi and uh thank you
everyone f... |
ec74fe65-6b33-4aef-b609-63a00e69eab6 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Anti-Alignments -- Measuring The Precision of Process Models and Event Logs
1 Introduction
---------------
Relating observed and modelled process behavior is the lion’s share of conformance checking [[9](#bib.bib9)]. Observed behavior is
often recorded in form of event logs, that store the footprints of process exe... |
074c06da-ea63-46c9-a3b0-73d7fdee2404 | trentmkelly/LessWrong-43k | LessWrong | Bay Winter Solstice 2024: Speech Auditions
Hello! I'm looking for community members to read speeches at the Bay Winter Solstice event this year. If you're interested, please email me at ozybrennan@gmail.com by the end of November 17 with the speeches you're potentially interested in. I'll ask you to record yourself re... |
85038a94-1373-414e-b816-a49516ad4af3 | trentmkelly/LessWrong-43k | LessWrong | Pondering how good or bad things will be in the AGI future
Yesterday I heard a podcast where someone said he hoped AGI would be developed in his lifetime. This confused me, and I realized that it might be useful - at least for me - to write down this confusion.
Consider that for some reasons - different history, diff... |
23db93fd-8f9b-421a-a342-d10dd162057e | trentmkelly/LessWrong-43k | LessWrong | Progress links and tweets, 2022-09-28
Announcements
* “Vast is developing artificial gravity space stations to expand humanity across the solar system”
* Derek Thompson and Ezra Klein are writing a book about abundance (@DKThomp)
* Foresight DAO to seed-fund “projects in longevity, nanotechnology, BCIs, private com... |
9aa12efc-6126-4923-b963-1f688f531248 | trentmkelly/LessWrong-43k | LessWrong | [Link]How to Achieve Impossible Career Goals (My manifesto on instrumental rationality)
Hey guys,
Don't normally post from my blog to here, but the latest massive post on goal achievement in 2015 has a ton that would be relevant to people here.
Some things that I think would be of particular interest to LWers:
... |
8d25ad2b-a307-4b76-9eec-687e5f50b3b1 | trentmkelly/LessWrong-43k | LessWrong | Does there exist an interactive reasoning map tool that lets users visually lay out claims, assign probabilities and confidence levels, and dynamically adjust their beliefs based on weighted influences between connected assertions?
I'm a professional investigator. My standard of evidence is "more likely than not." I t... |
66e01916-b2ef-45cd-aaa4-dc5b15b956b7 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Alignment Newsletter #49
Find all Alignment Newsletter resources [here](http://rohinshah.com/alignment-newsletter/). In particular, you can [sign up](http://eepurl.com/dqMSZj), or look through this [spreadsheet](https://docs.google.com/spreadsheets/d/1PwWbWZ6FPqAgZWOoOcXM8N_tUCuxpEyMbN1NYYC02aM/edit?usp=sharing) of al... |
5eab3d75-863e-4990-96fd-182544090730 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Announcement: AI alignment prize winners and next round
We (Zvi Mowshowitz, Vladimir Slepnev and Paul Christiano) are happy to announce that the [AI Alignment Prize](https://www.lesserwrong.com/posts/YDLGLnzJTKMEtti7Z/announcing-the-ai-alignment-prize) is a success. From November 3 to December 31 we received over 40 e... |
30a97eff-4094-4d89-92c1-579ed1824d3c | trentmkelly/LessWrong-43k | LessWrong | How tokenization influences prompting?
I was thinking about how prompt differs from training data in terms of tokenization. If i am to prompt with "solution:" as opposed to "solution: " it seems like it can influence the result, as in training data last token contain some information about next token. If there is toke... |
6b8807b0-1947-43de-967a-64b1edcf87e9 | StampyAI/alignment-research-dataset/blogs | Blogs | MIRI’s 2015 Summer Fundraiser!
This last year has been [pretty astounding](https://intelligence.org/2015/07/16/an-astounding-year/). Since its release twelve months ago, Nick Bostrom’s book *[Superintelligence](https://en.wikipedia.org/wiki/Superintelligence:_Paths,_Dangers,_Strategies#Reception)* has raised awareness... |
1bb63049-6849-4fe9-ad93-bc42e410a680 | trentmkelly/LessWrong-43k | LessWrong | Subsidizing Prediction Markets
Epistemic Status: Sadly Not Yet Subsidized
Robin Hanson linked to my previous post on prediction markets with the following note:
> .@TheZvi reviews what one needs for a working prediction market FUNDED BY TRADERS. If someone else will sponsor/subsidize them, a lot more becomes possibl... |
c95768cf-61d6-425d-ae0d-c51ba0f6074e | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Anthropic decision theory for self-locating beliefs
This is a link post to the "[Anthropic decision theory for self-locating beliefs](https://arxiv.org/pdf/1110.6437.pdf)" paper, with the abstract:
>
> This paper sets out to resolve how agents ought to act in the Sleeping Beauty problem and various related anthrop... |
da8290d3-b775-4511-83fb-b16194108b82 | trentmkelly/LessWrong-43k | LessWrong | Job Postings?
I often come across job postings that require a skill set - math, statistics, programming, etc - that might make them ideal for some LW readers. Since finding fun jobs for LWers seems like a good thing to do, I often think I should post these job profiles to the discussion section.
But nobody else has d... |
55073307-2db2-450d-b65d-7795a6b82367 | trentmkelly/LessWrong-43k | LessWrong | Where can I publish an article containing a list of intellectuals who publicly admitted their mistakes once proven wrong?
I’m writing a piece about (Public “I was wrong" admissions by intellectuals & thought leaders) that I’m planning to publish. It contains 6 people till now. Any tips on who would be interested in... |
bc8038c8-f216-4c58-90c9-c26006103f2d | trentmkelly/LessWrong-43k | LessWrong | Towards a mechanistic understanding of corrigibility
Acceptability
To be able to use something like relaxed adversarial training to verify a model, a necessary condition is having a good notion of acceptability. Paul Christiano describes the following two desiderata for any notion of acceptability:
1. "As long as th... |
8cefbb55-0f22-4015-8aee-d00d8ace7498 | trentmkelly/LessWrong-43k | LessWrong | When did humans become self-aware?
"If self-awareness emerged recently, this should show up in comparative linguistics..."
...
"Julian Jaynes should have fixed his date for the origin of consciousness to the origin of pronouns." |
efeb24a8-5b44-462a-aec1-1d1e0cb7c45c | trentmkelly/LessWrong-43k | LessWrong | Thoughts on AGI safety from the top
In this post, I'll summarize my views on AGI safety after thinking about it for two years while on the Research Scholars Programme at FHI -- before which I was quite skeptical about the entire subject. [1]
My goals here are twofold: (1) to introduce the problem from scratch in a wa... |
4d55790d-4d1c-4620-91e7-81f99bef9534 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | My first year in AI alignment
2022 was a pretty surprising year for me. In the beginning, I had just finished going through the [Recurse Center](https://www.recurse.com/) retreat, and was expecting to go back into software engineering. As it turned out, I spent almost all of it getting increasingly committed to workin... |
b3c53038-7c19-4ba1-8716-b73955ea5745 | trentmkelly/LessWrong-43k | LessWrong | Bounded surprise exam paradox
A small fun thing I came up with this morning.
The surprise exam paradox goes as follows: "you'll be given a surprise exam on one of the days next week, and you won't figure out the date of the exam on the previous day". Then you figure that the exam can't be on Sunday as it's the last d... |
5f5b23dd-eed7-46a8-9cdd-ab77f55994eb | trentmkelly/LessWrong-43k | LessWrong | [LINK] The errors, insights and lessons of famous AI predictions
The Journal of Experimental & Theoretical Artificial Intelligence has - finally! - published our paper "The errors, insights and lessons of famous AI predictions – and what they mean for the future":
Predicting the development of artificial intelligence... |
57ceda1b-20bc-4782-8c74-83d6d7f3ca4a | trentmkelly/LessWrong-43k | LessWrong | Swimming in Reasons
To a rationalist, certain phrases smell bad. Rotten. A bit fishy. It's not that they're actively dangerous, or that they don't occur when all is well; but they're relatively prone to emerging from certain kinds of thought processes that have gone bad.
One such phrase is for many reasons. For examp... |
571cb428-a34f-4dc6-8a51-cc64cfd730bf | trentmkelly/LessWrong-43k | LessWrong | Alignment versus AI Alignment
Financial status: This work is supported by individual donors and a grant from LTFF.
Epistemic status: This post contains many inside-view stories about the difficulty of alignment.
Thanks to Adam Shimi, John Wentworth, and Rob Miles for comments on this essay.
------------------------... |
ba3b80eb-000e-473d-8088-a657cfd939d4 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Why does deep and cheap learning work so well?
I Introduction
---------------
Deep learning works remarkably well, and has helped dramatically improve the state-of-the-art in areas ranging from speech recognition, translation and visual object recognition to drug discovery, genomics and automatic game playing LeCun... |
11ae480c-3e43-48b0-a33c-fee6e2c5024e | StampyAI/alignment-research-dataset/arxiv | Arxiv | Network Dissection:
Quantifying Interpretability of Deep Visual Representations
1 Introduction
---------------
| | | |
| --- | --- | --- |
| lamps in places net
| wheels in object net
| people in video net |
|
Refer to caption
|
Refer to caption
|
Refer to caption
|
Figure 1: Unit 13 in [[40](#bib... |
66abddbb-4c2b-4955-b09a-8fa3dfdef885 | trentmkelly/LessWrong-43k | LessWrong | Picture Frames, Window Frames and Frameworks
Some commenters on Noticing Frames were confused about what I actually meant by "frame." I defined it briefly as "different ways of seeing, thinking and communicating", but this was a bit vague.
I'm definitely using the word "frame" as a broad metaphor, rather than a concr... |
ce7acf6b-acfc-4884-9223-0089be0552f0 | trentmkelly/LessWrong-43k | LessWrong | Unsafe AI as Dynamical Systems
[Thanks to Valerie Morris for help editing this post.]
Overview
Large Language Models (LLMs) and their safety properties are often studied from the perspective of a single pass: what is the single next token the LLM will produce? But this is almost never how they are deployed. In contr... |
02a5bf01-bd14-4a92-9f50-93a922e9d92d | trentmkelly/LessWrong-43k | LessWrong | Book review: Pearl's Book of Why
Book review: The Book of Why, by Judea Pearl and Dana MacKenzie.
This book aims to turn the ideas from Pearl's seminal Causality) into something that's readable by a fairly wide audience.
It is somewhat successful. Most of the book is pretty readable, but parts of it still read like ... |
0ca60c3b-1b83-4e61-a35a-f93efc251854 | trentmkelly/LessWrong-43k | LessWrong | When EDT=CDT, ADT Does Well
Back in my first post about ADT, I conjectured that ADT might achieve sublinear regret relative to argmaxing for the true embedder, if some extra condition was imposed.
The Problem:
In the form of ADT stated in that post, the game of chicken is a counterexample to Conjecture 1. In short, ... |
21e66a53-ba40-4597-867d-64fd15acba00 | StampyAI/alignment-research-dataset/special_docs | Other | Inferring strategies from observations in long iterated Prisoner’s dilemma experiments.
[Download PDF](/articles/s41598-022-11654-2.pdf)
### Subjects
* [Computational science](/subjects/computational-science)
* [Scientific data](/subjects/scientific-data)
* [Social evolution](/subjects/social-evolution)
A... |
aeebdace-edee-42cd-a83f-20df500bd70b | StampyAI/alignment-research-dataset/arxiv | Arxiv | The 30-Year Cycle In The AI Debate
1 Introduction
---------------
The recent practical successes [[26](#bib.bib26)] of Artificial Intelligence (AI) programs of the Reinforcement Learning and Deep Learning varieties in game playing, natural language processing and image classification, are now calling attention to t... |
72f45321-8b58-488b-8ac2-ef80b1e9c8a7 | trentmkelly/LessWrong-43k | LessWrong | The Good News of Situationist Psychology
Part of the sequence: The Science of Winning at Life
In 1961, Stanley Milgram began his famous obedience experiments. He found that ordinary people would deliver (what they believed to be) excruciatingly painful electric shocks to another person if instructed to do so by an au... |
71e6946b-d3c8-4ab8-b275-71bfa15e26cc | trentmkelly/LessWrong-43k | LessWrong | The humility argument for honesty
|
7a665ee2-9436-41e5-aed7-a3d43430dbbb | trentmkelly/LessWrong-43k | LessWrong | Any writeups on GPT agency?
I'm searching for any articles that explore capabilities of GPT as an agent.
I don't seem to find a lot (or any for that matter). Am I bad at searching? |
0f0f6e0d-bef5-474d-ab7a-bd8c74e9b487 | trentmkelly/LessWrong-43k | LessWrong | Are there spaces for extremely short-form rationality content?
I'm experimenting with very short-form, compressed evocations of various mental adjustments. (Yes, I spend a lot of time on Twitter. Why do you ask?) For example, there's an adjunct to the idea of carving reality at the joints that could go something like ... |
f0627551-4e6b-46e3-ab30-ea0f64159f44 | trentmkelly/LessWrong-43k | LessWrong | Anthropic decision theory for self-locating beliefs
This is a link post to the "Anthropic decision theory for self-locating beliefs" paper, with the abstract:
> This paper sets out to resolve how agents ought to act in the Sleeping Beauty problem and various related anthropic (self-locating belief) problems, not thro... |
1a1971d7-8e1b-4ff7-8bbf-357f6c677652 | trentmkelly/LessWrong-43k | LessWrong | Research on repurposing filter products for masks?
A local organization is looking to produce masks for the hospital, and they reached out to me for guidance. Since this is outside of my expertise and I haven't been following all of the recent research, I'm passing the questions along here.
> We are looking for some ... |
d783ad05-0d19-47ff-9a86-74e4f0be56e8 | trentmkelly/LessWrong-43k | LessWrong | Reflective oracles as a solution to the converse Lawvere problem
(This post was originally published on Nov 30th 2017, and is 1 of 4 posts brought forwarded today as part of the AI Alignment Forum launch sequence on fixed points.)
1 Introduction
Before the work of Turing, one could justifiably be skeptical of the id... |
8ba7684f-859a-43a1-b125-1b273a982e97 | trentmkelly/LessWrong-43k | LessWrong | Why comparative advantage does not help horses
This post discusses what statements about comparative advantage say and what they do not say, and why comparative advantage does not save horses from getting sent to glue factories. It is only marginally about AI.
Eliezer Yudkowsky, in "The Sun is big, but superintellige... |
ba097c31-8184-4536-b558-95329fc58b71 | trentmkelly/LessWrong-43k | LessWrong | Three types of "should"
Note: This post is about "should" in how people think, in human psychology -- not about "should" in some deeper/broader philosophical sense that might apply to general agents.
I think when people talk about how they "should" do something, there's basically three different types of motivation/s... |
cef1064b-7f87-4977-8fe4-f0cf660a4570 | trentmkelly/LessWrong-43k | LessWrong | Short fic about uFAI
http://andrewhickey.info/2010/12/31/jeeves-and-the-singularity
Not sure if this is appropriate here, but I thought some of you might find this story I wrote amusing... |
9630848b-c934-4018-9af9-ded6957ac419 | trentmkelly/LessWrong-43k | LessWrong | Cognitive Impacts of Cocaine Use
This post summarizes The Long-term Effects of Cocaine Use on Cognitive Functioning: A Systematic Critical Review. The punchline is:
> The current evidence does not support the view that chronic cocaine use is associated with broad cognitive deficits. The view that cocaine users have ... |
541c2092-5a37-4d80-a99a-27062d911275 | trentmkelly/LessWrong-43k | LessWrong | Searching for Modularity in Large Language Models
Produced As Part Of The SERI ML Alignment Theory Scholars Program 2022 Under John Wentworth
See the Google Colab notebook to see the technical details of analysis that was done
Previous posts on Modularity investigated how one should search for and try to define modu... |
b4cc1551-f1f3-4fa9-85b8-5d98dc4e1489 | trentmkelly/LessWrong-43k | LessWrong | Cognitive Biases Affecting Self-Perception of Beauty
I wrote an article for mass consumption on the biases which are at play in a hot-button social issue, namely, how people feel about their beauty.
skepticexaminer.com/2016/05/dont-think-youre-beautiful/
and
intentionalinsights.org/why-you-dont-think-youre... |
a5868c73-ec9e-459a-a1f5-86b989df7466 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Australian Less Wrong Mega Meetup #2
Discussion article for the meetup : Australian Less Wrong Mega Meetup #2
WHEN: 17 July 2015 07:00:00PM (+1100)
WHERE: Barry Way, Jindabyne NSW 2627, Australia
We're doing it all again!
Following the success of last year's Mega-meetup, we're running another one - this t... |
f0bf092d-57ae-4fb2-89d2-1da5c3be64d4 | trentmkelly/LessWrong-43k | LessWrong | Attempts at Forwarding Speed Priors
This post summarizes research conducted under the mentorship of Evan Hubinger, and was assisted by collaboration with Pranav Gade, discussions with Adam Jermyn, and draft feedback from Yonadav Shavit.
Summary
1. Forwarding priors is a subproblem of deceptive alignment, because if... |
8c8cd707-9856-4be7-8260-665f4b70c77e | trentmkelly/LessWrong-43k | LessWrong | Thinking about maximization and corrigibility
Thanks in no small part to Goodhart's curse, there are broad issues with getting safe/aligned output from AI designed like "we've given you some function f(x), now work on maximizing it as best you can".
Part of the failure mode is that when you optimize for highly scorin... |
da9badd1-fad6-468b-aed1-4b1ba15e637e | trentmkelly/LessWrong-43k | LessWrong | Meetup : Tel Aviv: Choosing research topics
Discussion article for the meetup : Tel Aviv: Choosing research topics
WHEN: 26 June 2014 07:00:00PM (+0300)
WHERE: Derech Menachem Begin 7, 15th floor, Ramat Gan, Israel
At the next LessWrong meeting, on Thursday, June 26 at 19:00, we'll prepare for an upcoming MIRIx ses... |
e4f23e07-a379-4cbb-a20f-c6fa454b0863 | trentmkelly/LessWrong-43k | LessWrong | Jobs and internships available at the Centre for Effective Altruism
I work at the Centre for Effective Altruism, and thought some LessWrongers might be interested in applying for the jobs and internships that we're currently offering - or else know others who might be! CEA’s been discussed on LessWrong several times b... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.