~www_lesswrong_com | Bookmarks (657)

Published on September 22, 2024 2:03 PM GMTHi all! I'm creating a deep dive podcast episode about (the original) Leverage Research (for my podcast, Clearer Thinking). If you worked at (or currently work at) Leverage, or you have had experiences connected to Leverage, and you would like to either potentially:(1) do an interview for the podcast(2) talk to me totally off the record about...

Published on September 22, 2024 7:28 AM GMTexpectation calibrator: freeform draft, posting partly to practice lowering my own excessive standards.So, a few months back, I finally got around to reading Nick Bostrom's Superintelligence, a major player in the popularization of AI safety concerns. Lots of the argument was stuff I'd internalized a long time ago, reading the sequences and even living with AI alignment...

Published on September 22, 2024 3:20 AM GMTBackgroundFor background on YouCongress.com see this post by Hector Perez Arenas.I love this general concept, and have a lot of ideas for how this implementation could be expanded. I'm hoping that writing out some of my ideas might inspire someone to jump in an contribute code. I would myself if I didn't feel full to the brim...

Published on September 21, 2024 9:52 PM GMTIn small normal-form games, taking away an option from a player improves the payoff of that player usually <⅓ of the time, but improves the payoffs for the other player ½ of the time. The numbers depend on the size of the game; plotted here. There's been some discussion about the origins of paternalism. I believe that...

Published on September 21, 2024 4:54 PM GMTHey! I’m Kevin, an aspiring software engineer from Ghana. I’ve recently done the 80k 8 weeks career course and I’m excited to start earning-to-give. I’m looking for a mentor to help me land my first job and help me plan my career to maximise my expected donations! I’m in GMT +2. If you’d like to meet and figure out...

Published on September 21, 2024 4:30 PM GMTPreviously Alex Altair and I published a post on the applications of chaos theory, which found a few successes but mostly overhyped dead ends. Luckily the comments came through, providing me with an entirely different type of application: knowing you can’t, and explaining to your boss that you can’t. Knowing you can’t Calling a system chaotic rules...

Published on September 21, 2024 3:29 PM GMTWe (Flo & Rudolf) spent a month fleshing out the idea of an insurance-for-AI company. We talked to 15 people in the insurance industry, and did 20 customer interviews. We decided not to continue, but we think it’s still a very promising idea and that maybe someone else should do this. This post describes our findings.The ideaTheory of...

Published on September 20, 2024 6:30 PM GMTThis is a guest post by my student Ruiqi Zhong, who has some very exciting work defining new families of statistical models that can take natural language explanations as parameters. The motivation is that existing statistical models are bad at explaining structured data. To address this problem, we agument these models with natural language parameters, which can...

Published on September 20, 2024 6:20 PM GMTA very cool project, sponsored by the Chan Zuckerberg Initiative, was the Tabula Muris and the Tabula Muris Senis — a single-cell gene expression “atlas” of the mouse, and then mice at various ages from 3 months (young adulthood) to 27 months (near the end of their lifespan.)There’s a handy-dandy differential expression app that lets you see...

Published on September 20, 2024 5:21 PM GMTA very interesting problem is measuring something like general intelligence. I’m not going to delve deeply into this topic but simply want to draw attention to an idea that is often implied, though rarely expressed, in the framing of such a problem: the assumption that an "intelligence level," whatever it may be, corresponds to some inherent properties...

Published on September 20, 2024 5:17 PM GMTAnd how surprisingly untraumatizing it apparently is when mom feigns amnesia.I’m going to be honest here and say I just did the mental equivalent of a fist pump when my 5 year old responded to a prompt to pick any book in the book store with … The Handbook for Superheroes.Part 9.We are definitely starting at part...

Published on September 20, 2024 4:11 PM GMTTL;DR: After re-examining my earlier stance on collapsing the belief/knowledge distinction, I've come to realize through Wittgenstein's "meaning as use" that these terms hold distinct meanings based on their contextual applications in language. Embracing this understanding, I now see that preserving the distinction is essential for meaningful communication and continued dialogue.MotivationNine days ago, I made a post...

Published on September 20, 2024 1:13 PM GMTIntroductionI've spent the last few days going through every glitch token listed in the third SolidGoldMagikarp glitch token post, and was able to find the cause/source of almost every single one. This is the first in a series of posts in which I will explain the behaviors of these glitch tokens, the context in which they appear...

Published on September 20, 2024 11:12 AM GMTIn my previous post, I was talking about how inquisitive thinking – the search for the best possible arguments in support of a position – might render us ever more susceptible to our biases. I'd like to try to generalize, analogize and name that phenomenon as "argument overfitting", as well as discuss some of its consequences a...

Published on September 20, 2024 8:39 AM GMTPrevious post: How good are LLMs at doing ML on an unknown dataset?  A while back I ran some evaluation tests on GPT4o, Claude Sonnet 3.5 and Gemini advanced to see how good they were at doing machine learning on a completely novel, and somewhat unusual dataset. The data was basically 512 points in the 2D plane, and...

Published on September 20, 2024 2:30 AM GMT I had a great time playing Amherst last night with Cecilia, except for the keyboard gremlins. I just got a new keyboard (more on that later!) and it worked great when I tried it at home. At the dance, however, there were strange crackles. Even weirder, when I played the highest notes on the piano, I...

Published on September 18, 2024 6:33 PM GMTTL;DR:Alignment researchers have historically predicted that building safe advanced AI would necessarily incur a significant alignment tax compared to an equally capable but unaligned counterfactual AI. We put forward a case here that this prediction looks increasingly unlikely given the current ‘state of the board,’ as well as some possibilities for updating alignment strategies accordingly.IntroductionWe recently found that over one...

Published on September 18, 2024 1:50 PM GMTEpistemic status: prima facie unlikely in the usual framework, which I'll try to reframe. Corroborated by loads of empirical observations. YMMV, but if you've held some contrarian view in the past that you came to realize was wrong, this might resonate.In practical (and also not-so-practical) life, we often have to make a call as to which theory...

Published on September 18, 2024 2:05 PM GMTHi everyone! I’ve written an article I’m rather happy with on the history of endogenous growth models, and on the influence of intelligence on country level outcomes. As it is quite long, I will excerpt only a part — I sincerely hope you read the whole thing.https://nicholasdecker.substack.com/p/endogenous-growth-and-human-intelligence——————————————————ii. The History of Macroeconomic Growth ModelsMacro growth models start in...

Published on September 18, 2024 1:30 PM GMTThis post isn’t totally about the culture war topic du jour. Not at first.As with any other topic that soaks up angst like an ultra-absorbent sponge, I wonder how many have lost track of how we arrived here. Why are pronouns? Pronouns have always been meant to serve as a shortcut substitute reference for other nouns, and...

Published on September 18, 2024 2:05 AM GMTA year ago, I started trying to deliberate practice skills that would "help people figure out the answers to confusing, important questions." I experimented with Thinking Physics questions, GPQA questions, Puzzle Games , Strategy Games, and a stupid twitchy reflex game I had struggled to beat for 8 years[1]. Then I went back to my day job...

Published on September 17, 2024 11:48 PM GMTBeing able to quickly incorporate AI tools seems important, including for working on AI risk (people who disagree: there's a thread for doing so in the comments). But there are a lot of AI products and most of them suck.  Does anyone know a good source of reviews, or even just listing product features and naming obvious...