text stringlengths 300 320k | source stringlengths 52 154 |
|---|---|
# It’s Probably Not Lithium
*This post has been recorded as part of the LessWrong Curated Podcast, and can be listened to on* [*Spotify*](https://open.spotify.com/episode/7pndoqMFuCXH6dtB6tBk3z?si=abe97e4bb5e74de1)*,* [*Apple Podcasts*](https://podcasts.apple.com/us/podcast/lesswrong-curated-podcast/id1630783021)*,* [... | https://www.lesswrong.com/posts/7iAABhWpcGeP5e6SB/it-s-probably-not-lithium |
# Challenge: A Much More Alien Message
Since [this challenge](https://www.lesswrong.com/posts/dFFdAdwnoKmHGGksW/contest-an-alien-message) was solved pretty quickly, I thought I'd try to make a substantially more difficult program induction challenge.
### **The Challenge**
Find a short program which generates [this b... | https://www.lesswrong.com/posts/L3DX9Ptzdok5gJ63m/challenge-a-much-more-alien-message |
# Kurzgesagt – The Last Human (Youtube)
TLDW: Big Youtube channel Kurzgesagt released a video on the potential of humanity's future that's I think my favorite videos of theirs yet.
I have a pretty high bar for posting video content to LessWrong, but this video seemed better than most other videos in terms of capturin... | https://www.lesswrong.com/posts/xvHntkKzQfoFZfav3/kurzgesagt-the-last-human-youtube |
# Can We Align AI by Having It Learn Human Preferences? I’m Scared (summary of last third of Human Compatible)
*Epistemic status: I am just learning about alignment and just read Human Compatible. Below is a summary of the paradigm he outlines for aligning AI in the last third of his book, and the questions I have of... | https://www.lesswrong.com/posts/JjGh46oPtTe5xntHH/can-we-align-ai-by-having-it-learn-human-preferences-i-m |
# The table of different sampling assumptions in anthropics
[ ](https://twitter.com/anderssandberg/status/1539870473122713600) *“…. anthropic reasoning: complicated, error-prone,*
* looks a bit magical but once you start getting real evidence much *
* ... | https://www.lesswrong.com/posts/GBgumWfXWiqWxkt2K/the-table-of-different-sampling-assumptions-in-anthropics |
# Kevin Kelly's "103 Bits of Advice," Expanded
Kevin Kelly, who arguably has written one of the most influential essays regarding the internet and the creator economy, "[1,000 True Fans](https://kk.org/thetechnium/1000-true-fans/)", published "103 Bits of Advice I Wish I Had Known" on his 70th birthday. In the simple ... | https://www.lesswrong.com/posts/nGfajfToyjy8iLKWL/kevin-kelly-s-103-bits-of-advice-expanded |
# Will Capabilities Generalise More?
[Nate](https://www.alignmentforum.org/posts/GNhMPAWcfBCASy8e6/a-central-ai-alignment-problem-capabilities-generalization) and [Eliezer](https://www.alignmentforum.org/posts/uMQ3cqWDPHhjtiesc/agi-ruin-a-list-of-lethalities) (Lethality 21) claim that capabilities generalise further t... | https://www.lesswrong.com/posts/cq5x4XDnLcBrYbb66/will-capabilities-generalise-more |
# Limits to Legibility
From time to time, someone makes the case for why transparency in reasoning is important. The latest conceptualization is [*Epistemic Legibility*](https://www.lesswrong.com/posts/jbE85wCkRr9z7tqmD/epistemic-legibility) by Elizabeth, but the core concept is similar to [reasoning transparency](htt... | https://www.lesswrong.com/posts/4gDbqL3Tods8kHDqs/limits-to-legibility |
# Game Review: This Merchant Life
In a previous life, I used to be a professional game player. I still am hard at work on a new tradable blockchain card game, [Emergents](https://emergentstcg.com/), now in closed (but soon to be more open) beta, and I still love playing games.
This weekend I had the chance to take a ... | https://www.lesswrong.com/posts/KRHm9dxrdKGkaitER/game-review-this-merchant-life |
# Latent Adversarial Training
The Problem
===========
We’d like to train models to be robustly safe, even in environments that may fall well outside of the training distribution. Unfortunately all we get to work with is the training distribution, which makes ensuring robust generalization difficult.
Deception is an ... | https://www.lesswrong.com/posts/atBQ3NHyqnBadrsGP/latent-adversarial-training |
# Gradient hacking: definitions and examples
**EDIT: I have now shifted my terminology as follows (format: old term-->new term):**
* **"RL credit hacking"-->*****exploration hacking*****.**
* **"Gradient descent hacking"-->*****gradient hacking*****.**
* **"Gradient hacking"-->*****credit hacking*****.**
**Tha... | https://www.lesswrong.com/posts/EeAgytDZbDjRznPMA/gradient-hacking-definitions-and-examples |
# A summary of every Replacing Guilt post
1
=
I recently finished the [Replacing Guilt series](https://mindingourway.com/guilt/) (by Nate Soares).
I don't struggle much with guilt. I consider myself a happy and energetic person. I also spent several years studying psychology, and I'm familiar with many techniques f... | https://www.lesswrong.com/posts/M5hJhSzbGFJDgZvjv/a-summary-of-every-replacing-guilt-post |
# $500 bounty for alignment contest ideas
**\*Up to $500 for alignment contest ideas\***
A colleague and I are composing questions for an AI alignment talent search contest. We want to use (or come up with) a frame of the alignment problem that is accessible to smart high schoolers/college students and people wit... | https://www.lesswrong.com/posts/qpKYn4sF67ipposHB/usd500-bounty-for-alignment-contest-ideas |
# Cultivating And Destroying Agency
Something that really surprised me about high school is how little agency the majority of the students there had. This was even more pronounced among the students at the local magnet school, Academies of Loudoun; they were selected for on intelligence and work ethic, but generally h... | https://www.lesswrong.com/posts/rXX6vd2RrKHfePPfE/cultivating-and-destroying-agency |
# Formal Philosophy and Alignment Possible Projects
Context
=======
We ([Ramana](https://www.alignmentforum.org/users/ramana-kumar), [Abram](https://www.alignmentforum.org/users/abramdemski), [Josiah](https://josiahlopezwild.com), [Daniel](https://www.danielherrmann.ca)) are working together as part of [PIBBSS](https... | https://www.lesswrong.com/posts/fgAyy4gdDrbHFHjge/formal-philosophy-and-alignment-possible-projects |
# Covid 6/30/22: Vaccine Update Update
This week’s news is that the FDA advisory committee voted overwhelmingly to update the vaccine for Omicron, after a delay of only six months, which means they’ll get to deciding which way it should be updated Real Soon Now and then they can tell the pharma companies what requirem... | https://www.lesswrong.com/posts/83KuwSoruqiENMSxK/covid-6-30-22-vaccine-update-update |
# Abadarian Trades
*Spoilers for* [*planecrash (Book 2).*](https://www.glowfic.com/posts/5508)
> I propose an equal split of the gains from this trade, and will reject lesser splits with a probability corresponding to how disproportionately they reserve the gains for you, such that you can't actually do better by pre... | https://www.lesswrong.com/posts/uYycAnsumyGAAoDgv/abadarian-trades |
# Failing to fix a dangerous intersection
Over the last few years, many people have written about why America can't build things anymore (eg. [here](https://palladiummag.com/2022/06/09/why-america-cant-build/), although this is just one of hundreds of relevant essays). Ten years ago, when I was 21 and had just graduat... | https://www.lesswrong.com/posts/KdgJyHAr4AaPrwhf7/failing-to-fix-a-dangerous-intersection |
# [Linkpost] Solving Quantitative Reasoning Problems with
Language Models
A new paper from Google, in which they get a language model to solve some (of what to me reads as terrifyingly impressive) tasks which require quantitative reasoning skills. The abstract reads as follows:
> Language models have achieved remarka... | https://www.lesswrong.com/posts/JkKeFt2u4k4Q4Bmnx/linkpost-solving-quantitative-reasoning-problems-with |
# Quick survey on AI alignment resources
TLDR: In collaboration with CEA, an external collaborator is trying to determine the value of various AI alignment resources. If you are interested in or currently working on AI alignment, please fill out [**this survey**](https://www.guidedtrack.com/programs/qd2mn4k/run). We e... | https://www.lesswrong.com/posts/vNsFmrEqX74Ldem3H/quick-survey-on-ai-alignment-resources |
# The Track Record of Futurists Seems ... Fine
I've argued that the development of advanced AI could make this the [most important century](https://www.cold-takes.com/most-important-century/) for humanity. ... | https://www.lesswrong.com/posts/B2nBHP2KBGv2zJ2ew/the-track-record-of-futurists-seems-fine |
# Metacognition in the Rat
Cross-posted, as always, from [Putanumonit](https://putanumonit.com/2022/06/27/metacognition-in-the-rat/)
* * *
*Metacognition in the rat (n.)*
1. A brown rat’s (*Rattus norvegicus*) awareness of it’s own mental contents and ability to act on that awareness.
2. A rationalist’s (*Homo ac... | https://www.lesswrong.com/posts/CNCm3fcXQtXwYusQS/metacognition-in-the-rat |
# GPT-3 Catching Fish in Morse Code
*Mostly non-serious and slightly silly, with some potentially interesting bits for people who are into language models. *
*TLDR: The current version of GPT-3 has a strong tendency to encode mangled versions of a specific phrase when asked to write morse code in zero-shot situations... | https://www.lesswrong.com/posts/hDePh3KReBMNBJfzx/gpt-3-catching-fish-in-morse-code |
# Murphyjitsu: an Inner Simulator algorithm
**Epistemic status:** Firm
*The concepts underlying the Inner Simulator model and the related practical technique of Murphyjitsu are well-known and well-researched, including Kahneman’s S1/S2, mental simulation, and mental contrasting. Similarly, the problems that this unit... | https://www.lesswrong.com/posts/Htjbj8ystqc2zLtMX/murphyjitsu-an-inner-simulator-algorithm |
# Forecasts are not enough
I'll start the post with an example because it's the best way I know to explain what I want to talk about.
If you had to evaluate how well someone understands macroeconomics but you yourself didn't have the mental tools to do so on the basis of their arguments, how would you go about doing ... | https://www.lesswrong.com/posts/uDz4ydD8dZBdm9PgZ/forecasts-are-not-enough |
# Cryonics-adjacent question
(This question involves my somewhat bizarre "[mind backup](https://www.lesswrong.com/posts/gZzLaKGuaMqYwmHeJ/what-i-m-about)" project, that was inspired by Rudy Rucker's "Lifebox" idea.)
Unfortunately, I made some HUGE blunders last year involving my life savings and the market (bonds, th... | https://www.lesswrong.com/posts/yhhGJYo2jJqEwSgos/cryonics-adjacent-question |
# Selection processes for subagents
*This post consists of rough thoughts on possible training processes that select for multi-agent architectures, or “subagents”, in neural networks. I deliberately ignore *[*delegative subagents that are instantiated by mesa-optimizers*](https://www.lesswrong.com/posts/8RCCMStERhfkYZ... | https://www.lesswrong.com/posts/d4hw4FBX9YXHGFBWQ/selection-processes-for-subagents |
# Looking back on my alignment PhD
*This post has been recorded as part of the LessWrong Curated Podcast, and can be listened to on* [*Spotify*](https://open.spotify.com/episode/5UY1LrzUakTbs8LsL0ld1s?si=adfa86cec2c4409a)*,* [*Apple Podcasts*](https://podcasts.apple.com/us/podcast/looking-back-on-my-alignment-phd-b... | https://www.lesswrong.com/posts/2GxhAyn9aHqukap2S/looking-back-on-my-alignment-phd |
# AI safety university groups: a promising opportunity to reduce existential risk
[*Cross-posted from the Effective Altruism Forum*](https://forum.effectivealtruism.org/posts/bL2CWihhaSjpqk8QD/ai-safety-university-groups-a-promising-opportunity-to)
*Summary*: AI safety university groups[^vmyfd7tkhp] are a promising w... | https://www.lesswrong.com/posts/F6D5r2oq9CCnEBhWo/ai-safety-university-groups-a-promising-opportunity-to |
# One is (almost) normal in base π
See previous posts on my blog if you’re confused:
**Part 1:** [Normal numbers in base ten](https://adam.scherlis.com/2022/06/30/non-unique-decimals-0-999-and-normal-numbers-in-base-ten/)
**Part 2:** [Numbers in base φ and π](https://adam.scherlis.com/2022/06/30/numbers-in-base-%cf... | https://www.lesswrong.com/posts/k2FcbPvzbBxkc25va/one-is-almost-normal-in-base-p |
# How to Navigate Evaluating Politicized Research?
I want to look into the research literature on an issue that has unfortunately become extremely politicized and controversial. (It isn't gun ownership, but let's say for the sake of argument that it is.) If I do a search for, say, "effects of gun ownership on crime", ... | https://www.lesswrong.com/posts/vZfmwjgmiYTKiwdMM/how-to-navigate-evaluating-politicized-research |
# Meiosis is all you need
Cross-posted from [my Substack.](https://denovo.substack.com/p/meiosis-is-all-you-need)
*Note: this post is still a rough draft but I think this idea is important enough to publish it anyway. I intend to revise it in the future. Also if you are a biologist who wants to collaborate with me on... | https://www.lesswrong.com/posts/zWKA48izeoq5HcnkD/meiosis-is-all-you-need |
# Open & Welcome Thread - July 2022
If it’s worth saying, but not worth its own post, here's a place to put it.
If you are new to LessWrong, here's the place to introduce yourself. Personal stories, anecdotes, or just general comments on how you found us and what you hope to get from the site and community are invite... | https://www.lesswrong.com/posts/zXRjyLyod7PDuzRf8/open-and-welcome-thread-july-2022 |
# Safetywashing
In southern California there’s a two-acre butterfly preserve owned by the oil company Chevron. They spend little to maintain it, but many millions on television advertisements featuring it as evidence of their environmental stewardship.[^9m6ozmbv1h]
Environmentalists have a word for behavior like this... | https://www.lesswrong.com/posts/xhD6SHAAE9ghKZ9HS/safetywashing |
# What Is The True Name of Modularity?
TL;DR
=====
Modularity seems like an important feature of neural networks, but there is currently no canonical way of properly defining or measuring it which is properly theoretically motivated and doesn’t break down in some cases - in other words, we haven’t yet found a [True N... | https://www.lesswrong.com/posts/TTTHwLpcewGjQHWzh/what-is-the-true-name-of-modularity |
# Trends in GPU price-performance
Executive Summary
=================
Using a dataset of 470 models of graphics processing units (GPUs) released between 2006 and 2021, we find that the amount of floating-point operations/second per $ (hereafter FLOP/s per $) **doubles every ~2.5 years**. For top GPUs, we find a slowe... | https://www.lesswrong.com/posts/c6KFvQcZggQKZzxr9/trends-in-gpu-price-performance |
# Limerence Messes Up Your Rationality Real Bad, Yo
There’s a pretty basic rationality fact that I don’t see talked about much on LW despite it’s obvious relevance. So I am here to write The Canonical Rationality Post on the topic:
[Limerence](https://en.wikipedia.org/wiki/Limerence) (aka "falling in love"[^edgflkfy0... | https://www.lesswrong.com/posts/Jk9yMXpBLMWNTFLzh/limerence-messes-up-your-rationality-real-bad-yo |
# Who is this MSRayne person anyway?
I've been on this site as a lurker for a few years now, and started interacting more heavily the past month or so. I've been reluctant to engage much, because, to be honest, I'm a very unusual person - even by the standards of LessWrong - and I don't feel like I fit in or have any ... | https://www.lesswrong.com/posts/44psWMpmdxBwmzAYK/who-is-this-msrayne-person-anyway |
# Reframing the AI Risk
**Follow-up to:** [Reshaping the AI Industry: Straightforward Appeals to Insiders](https://www.lesswrong.com/posts/mF8dkhZF9hAuLHXaD/reshaping-the-ai-industry#3_1__Straightforward_Appeals_to_Insiders)
* * *
Introduction
------------
[The central issue](https://www.lesswrong.com/posts/Rkxj7TF... | https://www.lesswrong.com/posts/m3fyWQgCcFwro5KQh/reframing-the-ai-risk |
# Paper: Forecasting world events with neural nets
**Paper authors:** Andy Zou, Tristan Xiao, Ryan Jia, Joe Kwon, Mantas Mazeika, Richard Li, Dawn Song, Jacob Steinhardt, Owain Evans, Dan Hendrycks
This post contains the [paper’s](https://arxiv.org/abs/2206.15474) abstract and excerpts from the paper (with slight mod... | https://www.lesswrong.com/posts/ixiCECkFFguFcHWFa/paper-forecasting-world-events-with-neural-nets |
# Minerva
Google Research's new AI tackles natural language math problems and handily outperforms the SOTA[^z9ujz7hobka]. It is a pre-trained PaLM [^jhan4auwcm]finetuned on some maths datasets (which use LaTeX) composed of maths webpages and Arxiv papers (38.5B tokens). The three models trained were as follows.
Many people in the AI alignment space have heard of AI safety via debate - check out [episode 6 of AXRP](https://axrp.net/episode/2021/04/08/epi... | https://www.lesswrong.com/posts/bLr68nrLSwgzqLpzu/axrp-episode-16-preparing-for-debate-ai-with-geoffrey-irving |
# Agenty AGI – How Tempting?
\[Epistemic status: mostly writing to clarify my intuitions, with just a
few weak attempts to convince others. It\'s no substitute for reading
Drexler\'s writings.\]
I\'ve been struggling to write more posts relating to Drexler\'s vision
for AI (hopefully to be published soon), and i... | https://www.lesswrong.com/posts/86CxnQzPdFoDhdMXh/agenty-agi-how-tempting |
# [Linkpost] Existential Risk Analysis in Empirical Research Papers
I’ve noted before that it can be difficult to separate progress in safety from progress in capabilities. However, doing so is important, as we want to ensure that we are making differential progress on safety, rather than just advancing safety as a co... | https://www.lesswrong.com/posts/5rNCGP8deEBjedCmH/linkpost-existential-risk-analysis-in-empirical-research |
# Five views of Bayes' Theorem
I think rearranging Bayes' Theorem sheds light on it in interesting ways.
P(A|B) = P(B|A) P(A)/P(B)
-------------------------
This is the most common form: you want to know the conditional probability of A given B, but what you actually know is the probability of B given A (and your pr... | https://www.lesswrong.com/posts/RafNNHbnPvGDHydYn/five-views-of-bayes-theorem |
# Deontological Evil
It surprised Doctor Susan Connor how quickly she had gotten used to her job. At first she had qualms working for an organization literally called "Effective Evil" (with two capital 'E's). But her job just gave her so much autonomy, complexity and reward. Also nemeses.
"I'll never talk," said the ... | https://www.lesswrong.com/posts/eRDF3FjBgHnJHrKCC/deontological-evil |
# Literature recommendations July 2022
It happens frequently, that I'm researching a topic and don't really know which sources to read to get a good grasp of the topic. I expect that many fellow aspiring rationalists feel the same.
This thread is a place to list topics for which you are looking for reading recommend... | https://www.lesswrong.com/posts/kPETbhurC2ynstaq6/literature-recommendations-july-2022 |
# Naive Hypotheses on AI Alignment
Apparently doominess works for my brain, cause Eliezer Yudkowsky’s [AGI Ruin: A List of Lethalities](https://www.lesswrong.com/posts/uMQ3cqWDPHhjtiesc/agi-ruin-a-list-of-lethalities) convinced me to look in to AI safety. Either I’d find out he’s wrong, and there is no problem. Or he’... | https://www.lesswrong.com/posts/ubdp8qAL8Gfki2pYo/naive-hypotheses-on-ai-alignment |
# Trigger-Action Planning
**Epistemic status:** Established and confirmed
*There has been a tremendous amount of research on “implementation intentions” since their development by psychologist Peter Gollwitzer in the late 1990’s. A meta- analysis of 94 studies involving 8461 participants found that interventions usin... | https://www.lesswrong.com/posts/W5HcGywyPoDDdJtbz/trigger-action-planning |
# 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 misaligne... | https://www.lesswrong.com/posts/HoQ8WaEHXdkaMbpzx/can-we-achieve-agi-alignment-by-balancing-multiple-human |
# Sexual self-acceptance
I have noticed very often that I can't predict how my brain reacts towards women. I often expect that nothing special will happen, compared to talking to males. However very often sexual thoughts come to mind that seems out of place to me. Normally they are very short (1-5s) and visual, but wi... | https://www.lesswrong.com/posts/GxautDM9Q3cWKebPX/sexual-self-acceptance |
# SOMA - A story about Consciousness
SOMA has a great story. I like it so much that I made a story video compilation about it, containing only the best things. [The video](https://www.youtube.com/watch?v=Kp_Eroq5cOo) actually presents a coherent story, that somebody can enjoy without any prior knowledge, even though ~... | https://www.lesswrong.com/posts/fePzDFGCHiScReYQa/soma-a-story-about-consciousness |
# Nature abhors an immutable replicator... usually
*Epistemic status:* Popped into my head yesterday and I am trying to clarify my reasoning about it out loud. I expect that I am overestimating the universality of this principle to at least some degree, and there is some chance that it's altogether wrong, but I think ... | https://www.lesswrong.com/posts/hti2q9AA8efceAoAb/nature-abhors-an-immutable-replicator-usually |
# Toni Kurz and the Insanity of Climbing Mountains
*This post has been recorded as part of the LessWrong Curated Podcast, and can be listened to on* [*Spotify*](https://open.spotify.com/episode/0An7FpsWiNGFHzyxKb9d4W)*,* [*Apple Podcasts*](https://podcasts.apple.com/us/podcast/toni-kurz-and-the-insanity-of-climbing-mo... | https://www.lesswrong.com/posts/J3wemDGtsy5gzD3xa/toni-kurz-and-the-insanity-of-climbing-mountains |
# Five routes of access to scientific literature
**Tl;dr:** It seems like you can get access to a large amount of scientific literature on [DeepDyve](https://www.deepdyve.com/) for $500. This might be enough to meet the needs of researchers outside the university system.
> The eleventh virtue is scholarship. Study ma... | https://www.lesswrong.com/posts/TCTtaFPqbMrQhttCD/five-routes-of-access-to-scientific-literature |
# Decision theory and dynamic inconsistency
Here is my current take on decision theory:
* When making a decision after observing X, we should condition (or causally intervene) on statements like “My decision algorithm outputs Y after observing X.”
* Updating seems like a _description_ of something you do when mak... | https://www.lesswrong.com/posts/9W4TQvixiQjpZmzrx/decision-theory-and-dynamic-inconsistency |
# New US Senate Bill on X-Risk Mitigation [Linkpost]
Two US Senators have introduced a bipartisan bill specifically focused on x-risk mitigation, including from AI. From the [post on Senate.gov](https://www.hsgac.senate.gov/media/majority-media/peters-introduces-bipartisan-bill-to-ensure-federal-government-is-prepared... | https://www.lesswrong.com/posts/bHBmmkdwsmjNwp94K/new-us-senate-bill-on-x-risk-mitigation-linkpost |
# A compressed take on recent disagreements
An underlying generator of many recent disagreements appears to be differing locations of the disagreeing individuals on the [Yudkowsky-Hanson spectrum](http://intelligence.org/files/AIFoomDebate.pdf). That is, differences in individuals' distributions over the background va... | https://www.lesswrong.com/posts/cwz5mRM5KvtdXftzk/a-compressed-take-on-recent-disagreements |
# AI Forecasting: One Year In
Last August, my research group [created a forecasting contest](https://bounded-regret.ghost.io/ai-forecasting/) to predict AI progress on four benchmarks. Forecasts were asked to predict state-of-the-art performance (SOTA) on each benchmark for June 30th 2022, 2023, 2024, and 2025. It’s n... | https://www.lesswrong.com/posts/CJw2tNHaEimx6nwNy/ai-forecasting-one-year-in |
# Remaking EfficientZero (as best I can)
Introduction
------------
When I first heard about EfficientZero, I was amazed that it could learn at a sample efficiency comparable to humans. What's more, it was doing it without the gigantic amount of pre-training the humans have, which I'd always felt made comparing sample... | https://www.lesswrong.com/posts/bPa6AzRgGZGmxbq6n/remaking-efficientzero-as-best-i-can |
# Is General Intelligence "Compact"?
Epistemic Status
----------------
Exploratory and unconfident, but I believe important.
Acknowledgements
----------------
I'm grateful to "JustisMills", "sigilyph", "tomcatfish" and others for their valuable feedback on drafts of this post.
* * *
Introduction
============
By ... | https://www.lesswrong.com/posts/J9XecqtiujawmDnmr/is-general-intelligence-compact |
# My Most Likely Reason to Die Young is AI X-Risk
Introduction
------------
One of my pet peeves is people calling AI Safety a longtermist cause area. It is [entirely](https://forum.effectivealtruism.org/posts/KDjEogAqWNTdddF9g/long-termism-vs-existential-risk) [reasonable](https://forum.effectivealtruism.org/posts/r... | https://www.lesswrong.com/posts/MYCutshqJwTbHfE3W/my-most-likely-reason-to-die-young-is-ai-x-risk |
# Book Review: The Righteous Mind
I - Intuitions And Reason
-------------------------
They say cognitive biases are what *other *people have. In that vein, “The Righteous Mind” by Jonathan Haidt is a book about how *other* people think. Specifically, how they think about morality and make moral judgments and choices.... | https://www.lesswrong.com/posts/iBDkwir2FsG4DAiDh/book-review-the-righteous-mind-3 |
# Anthropic's SoLU (Softmax Linear Unit)
Anthropic continues their [Transformer Circuits Thread](https://transformer-circuits.pub/). In previous work they were unable to make much progress interpreting MLP layers. This paper is focused on addressing that limitation.
High-level Takeaways
====================
> In thi... | https://www.lesswrong.com/posts/PTuqkWYsioH2rhzwJ/anthropic-s-solu-softmax-linear-unit |
# Procedural Executive Function, Part 1
To most people, Executive Function is a confusing, mysterious thing that only really comes to attention through *dys*function, particularly mental health disorders like ADHD or depression. Questions of how it works, or how to get more of it, are treated as similar to questions o... | https://www.lesswrong.com/posts/qXcLkF3nqQyBX3gzj/procedural-executive-function-part-1 |
# Benchmark for successful concept extrapolation/avoiding goal misgeneralization
If an AI has been trained on data about adults, what should it do when it encounters a child? When an AI encounters a situation that its training data hasn’t covered, there’s a risk that it will incorrectly generalize from what it has bee... | https://www.lesswrong.com/posts/DiEWbwrChuzuhJhGr/benchmark-for-successful-concept-extrapolation-avoiding-goal |
# ITT-passing and civility are good; "charity" is bad; steelmanning is niche
*This post has been recorded as part of the LessWrong Curated Podcast, and can be listened to on* [*Spotify*](https://open.spotify.com/episode/7kbiGrZFQmMnc7o9o1MwIy)*,* [*Apple Podcasts*](https://podcasts.apple.com/us/podcast/itt-passing-and... | https://www.lesswrong.com/posts/MdZyLnLHuaHrCskjy/itt-passing-and-civility-are-good-charity-is-bad |
# [AN #172] Sorry for the long hiatus!
Listen to this newsletter on [The Alignment Newsletter Podcast](http://alignment-newsletter.libsyn.com/).
Alignment Newsletter is a publication with recent content relevant to AI alignment. Find all Alignment Newsletter [resources here](http://rohinshah.com/alignment-newslet... | https://www.lesswrong.com/posts/rxsdSgnZrgYWc2XAp/an-172-sorry-for-the-long-hiatus |
# Assorted thoughts about abstraction
_Epistemic status: Exploratory + I'm not an expert here. Still, I'm reasonably confident that I'm onto something._
Let's talk about levels of abstraction. I have an example that I think would be helpful.
It comes from the first computer science lecture that I ever attended. I t... | https://www.lesswrong.com/posts/4XmBwf4LpSumqvGZA/assorted-thoughts-about-abstraction |
# Goal Factoring
**Epistemic status**: Anecdotally strong
*The goal factoring technique as taught by CFAR is not derived from any particular body of psych research, but is instead a straightforward and general application of the principle of reductionism. It was developed and refined through iteration, and has been u... | https://www.lesswrong.com/posts/ZHWiCM4QmX8WwYajH/goal-factoring-1 |
# When is it appropriate to use statistical models and probabilities for decision making ?
I enjoy reading rationalist and effective altruist blogs. Members of the community usually back their arguments with data and/or other evidence and tend to be scientifically literate. And although no one is a perfect, detached b... | https://www.lesswrong.com/posts/WhTChcvmv45hcuPXy/when-is-it-appropriate-to-use-statistical-models-and |
# The curious case of Pretty Good human inner/outer alignment
I [have ](https://www.lesswrong.com/posts/HYERofGZE6j9Tuigi/inner-alignment-failures-which-are-actually-outer-alignment)[been convinced](https://astralcodexten.substack.com/p/deceptively-aligned-mesa-optimizers) to [believe](https://www.lesswrong.com/posts/... | https://www.lesswrong.com/posts/A5XTgYmGnqEDnJdzJ/the-curious-case-of-pretty-good-human-inner-outer-alignment |
# A Pattern Language For Rationality
There's a lens to looking at the rationality project that I've been finding enlivening recently, and I think it's reached the point where more eyes and hands might be useful, while not being anywhere near complete yet. First, some background.
Christopher Alexander was a designer a... | https://www.lesswrong.com/posts/eJcYxpwESvMRYGiyo/a-pattern-language-for-rationality |
# Norfolk Rationalists Weekly Wednesday Meetup @ Fair Grounds
Hello!
Please come on out to Fair Grounds for good conversation, good food & drink, and a lovely atmosphere :)
When: We meet each Wednesday from 17:00-19:30 EDT.
Where: Fair Grounds, an excellent coffee shop located at 806 Baldwin Ave # 2, Norfolk, VA 23... | https://www.lesswrong.com/events/JPtZoC7xExrapNhr7/norfolk-rationalists-weekly-wednesday-meetup-fair-grounds-5 |
# My vision of a good future, part I
A future where everyone is okay
-------------------------------
I grew up with the future feeling pretty uncertain. God was supposed to return at some point, but maybe there were going to be The End Times before that? Like, a great persecution of the Christians, and maybe somethin... | https://www.lesswrong.com/posts/sumqh9GWd7pf3akqe/my-vision-of-a-good-future-part-i |
# Introducing the Fund for Alignment Research (We're Hiring!)
*Cross-posted to the* [*EA Forum*](https://forum.effectivealtruism.org/posts/gNHjEmLeKM47FDdqM/introducing-the-fund-for-alignment-research-we-re-hiring-1)
The [Fund for Alignment Research](https://alignmentfund.org/) (FAR) is hiring research engineers and ... | https://www.lesswrong.com/posts/hGE3Pcc7qmK75bjhc/introducing-the-fund-for-alignment-research-we-re-hiring |
# Predicting Parental Emotional Changes?
Before I became a parent I cared about children in general, intellectually, in the same way that I care about people in general: a general sense that I want good things to happen to them. Over the first few days of Lily's life, however, I noticed I cared for her deeply and emot... | https://www.lesswrong.com/posts/2eQCQGpoRrEafpfCF/predicting-parental-emotional-changes |
# Carrying the Torch: A Response to Anna Salamon by the Guild of the Rose
In a recent comment and [followup post](https://www.lesswrong.com/posts/B9kP6x5rpmuCzpfWb/comment-reply-my-low-quality-thoughts-on-why-cfar-didn-t-get), Anna Salamon described some of the rocks upon which the Center for Applied Rationality has r... | https://www.lesswrong.com/posts/XLbDQL2qYi9FDozvL/carrying-the-torch-a-response-to-anna-salamon-by-the-guild |
# Deep neural networks are not opaque.
In his [List of Lethalities](https://www.lesswrong.com/posts/uMQ3cqWDPHhjtiesc/agi-ruin-a-list-of-lethalities), Eliezer writes that "matrices \[in neural networks\] are opaque". But I think our understanding of how neural networks actually work has improved significantly last yea... | https://www.lesswrong.com/posts/9D9GyvKDzgibCGqmq/deep-neural-networks-are-not-opaque |
# Tarnished Guy who Puts a Num on it
Why Elden Ring is great, and how not to die binomially.
Cross-posted from [Putanumonit](https://putanumonit.com/2022/04/21/tarnished-guy-who-puts-a-num-on-it/)
* * *
One reason I haven’t written as much in the last two months is that I’ve been playing a lot of [**Elden Ring*... | https://www.lesswrong.com/posts/K5DZyGg2fwf5EhqzG/tarnished-guy-who-puts-a-num-on-it |
# Outer vs inner misalignment: three framings
A core concept in the field of AI alignment is a distinction between two types of misalignment: outer misalignment and inner misalignment. Roughly speaking, the outer alignment problem is the problem of specifying an reward function which captures human preferences; and th... | https://www.lesswrong.com/posts/poyshiMEhJsAuifKt/outer-vs-inner-misalignment-three-framings-1 |
# Some Adventures of a Curious Richard Feynman
This is cross-posted [from my blog](https://www.daltonmabery.com/essays/adventures-of-a-curious-richard-feynman).
* * *
****Sitting around a campfire listening to stories passed on from generations brings with it a feeling of wonder and joy for the adventures told fro... | https://www.lesswrong.com/posts/Hg2MxEdFsuTvHgCXp/some-adventures-of-a-curious-richard-feynman |
# Confusions in My Model of AI Risk
A lot of [the reason I am worried about AI](https://www.lesswrong.com/posts/k8hvGAJWSKAeHwpnJ/why-i-m-worried-about-ai) comes from the development of optimizers that have goals which don’t align with what humans want. However, I am also pretty confused about the specifics here, espe... | https://www.lesswrong.com/posts/aZ2kHQtpTHu3FeguQ/confusions-in-my-model-of-ai-risk |
# Principles for Alignment/Agency Projects
"John, what do you think of this idea for an alignment research project?"
I get questions like that fairly regularly. How do I go about answering? What principles guide my evaluation? Not all of my intuitions for what makes a project valuable can easily be made legible, but ... | https://www.lesswrong.com/posts/A7GeRNLzuFnhvGGgb/principles-for-alignment-agency-projects |
# Race Along Rashomon Ridge
### *Produced As Part Of The SERI ML Alignment Theory Scholars Program 2022 Research Sprint Under* [*John Wentworth*](https://www.lesswrong.com/users/johnswentworth)
Two Deep Neural Networks with wildly different parameters can produce equally good results. Not only can a tweak to paramete... | https://www.lesswrong.com/posts/Hna2P8gcTyRgNDYBY/race-along-rashomon-ridge |
# Wealth as a source of technological stagnation?
*(This is an essay I wrote last year, but hadn't published. I'm posting it here because some friends liked it and there's no particular reason to keep it private.)*
The [World Values Survey](https://www.worldvaluessurvey.org/WVSOnline.jsp) is a very comprehensive set ... | https://www.lesswrong.com/posts/N6uNzcA7GGuTqji7n/wealth-as-a-source-of-technological-stagnation |
# LessWrong Meetup - Changing our Minds
One of the core marks of the rationalist is the ability to change one's mind. In the workshops, we will explore how we changed our minds in the past.
We will share insights with each other and look at which ways of changing our minds work for us.
16:00-16:15 Arriving and chat... | https://www.lesswrong.com/events/bNHsPkinfEqtD4bWh/lesswrong-meetup-changing-our-minds |
# When Giving People Money Doesn’t Help
I’ve been thinking a bunch about [this study](https://twitter.com/cblatts/status/1526532090430377989) ([direct link](https://t.co/w35kxWMsLf)) of CBT-style techniques used to try and get criminals onto a better life path in Liberia that showed what looks like spectacular results... | https://www.lesswrong.com/posts/nvyucWCsKkB5mFNYQ/when-giving-people-money-doesn-t-help |
# Babysitting as Parenting Trial?
Julia recently posted an [interview with Kat Woods](https://juliawise.net/interview-with-kat-woods-decision-making-about-having-kids/) on how Kat has thought through decisions around having kids. In her 20s she started strongly wanting to:
> I didn't want kids when I was younger, and... | https://www.lesswrong.com/posts/9iSgQMooPxNtPoTbk/babysitting-as-parenting-trial |
# Covid 7/7/22: Paxlovid at the Pharmacy
Pharmacists are going to be allowed to prescribe Paxlovid, while doctors continue to be confused by the guideline of ‘you can give this life-saving medicine that cures disease to the people who are sick with that disease.’ The news on the vaccine update front is good. Cases and... | https://www.lesswrong.com/posts/naJBjFY6wKaWuA7C8/covid-7-7-22-paxlovid-at-the-pharmacy |
# Consider Multiclassing
One approach to work is to specialize. The programmer who does lots of LeetCode in their spare time, keeps a portfolio up to date with side projects, and regularly posts answers on StackOverflow. The researcher who thinks about the same subject for the vast majority of their waking hours every... | https://www.lesswrong.com/posts/emmDNESMQJhoFodfr/consider-multiclassing |
# Genders Discrimination
Whales are queering the fish-mammal binary
Cross-posted from [Putanumonit](https://putanumonit.com/2022/05/02/genders-discrimination/)
* * *
Categories in your head
-----------------------
In physical reality, light with wavelength between 380 and 700 nm passes through your eyeball and act... | https://www.lesswrong.com/posts/hAtvD9wWoq6GQLXEz/genders-discrimination |
# Aversion Factoring
**Epistemic status**: Anecdotally strong
*The aversion factoring technique is not derived from any particular body of psych research, though it draws lightly on trigger-affect patterns and exposure therapy and takes advantage of the framework of reductionism. It was primarily developed for person... | https://www.lesswrong.com/posts/dvzwqLbpSK2nD8Yvn/aversion-factoring |
# Cooperation with and between AGI\'s
\[This post is primarily aimed at a Foresight Institute audience. I
don\'t know whether the average LW reader will learn much from it.\]
This post is mostly a response to the Foresight Institute\'s book
[Gaming the
Future](https://foresightinstitute.substack.com/p/coming-soo... | https://www.lesswrong.com/posts/eS23rutFfLnLvexXK/cooperation-with-and-between-agi-s-1 |
# Human values & biases are inaccessible to the genome
*Related to Steve Byrnes’ *[*Social instincts are tricky because of the “symbol grounding problem.”*](https://www.lesswrong.com/s/HzcM2dkCq7fwXBej8/p/5F5Tz3u6kJbTNMqsb#13_2_2_Claim_2__Social_instincts_are_tricky_because_of_the__symbol_grounding_problem_) *I wouldn... | https://www.lesswrong.com/posts/CQAMdzA4MZEhNRtTp/human-values-and-biases-are-inaccessible-to-the-genome |
# Safety considerations for online generative modeling
**Summary:** the [online decision transformer](https://arxiv.org/pdf/2202.05607.pdf) is a recent approach to creating agents in which a decision transformer is pre-trained offline (as usual) before producing its own trajectories which are fed back into the model i... | https://www.lesswrong.com/posts/BMfNu82iunjqKyQA9/safety-considerations-for-online-generative-modeling |
# How to Become a World Historical Figure (Péladan's Dream)
I. Introduction
---------------
*.*
The concept of "strategic clarity" has recently become increasingly important to how I think. It ... | https://www.lesswrong.com/posts/oXkSuqj5WfSY8CfYH/in-search-of-strategic-clarity |
# Exposition as science: some ideas for how to make progress
In a [previous post](https://www.lesswrong.com/posts/BddavsuaHLAhkH45Z/how-to-get-people-to-produce-more-great-exposition-some), I argued for treating exposition as a scientific field of inquiry as the best main strategy to use to get people to create more g... | https://www.lesswrong.com/posts/J3Edt2CLcXPHQYSXo/exposition-as-science-some-ideas-for-how-to-make-progress |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.