text stringlengths 300 320k | source stringlengths 52 154 |
|---|---|
# Finding the Central Limit Theorem in Bayes' rule
I feel like there's something in the basic Bayesian update formula that suggests that, as the amount of input data grows, almost every Bayesian posterior will tend toward a Gaussian. This post is me trying to scrape at that vague intuition and get a handle on it. I'm ... | https://www.lesswrong.com/posts/bEyn9h3SGeH59vnDM/finding-the-central-limit-theorem-in-bayes-rule |
# Should PAXLOVID be given preventively?
An effective antiviral against COVID-19 seems to have a potential to end the pandemic if we give it to the whole population. Has anyone evaluated whether PAXLOVID has the potential to do the job? How is the side effect profile when thinking about giving it preventively? | https://www.lesswrong.com/posts/hB6hrHxLbHKRefqKT/should-paxlovid-be-given-preventively |
# I currently translate AGI-related texts to Russian. Is that useful?
I live in Saint-Petersburg, Russia, and I want to do something useful to AI-Alignment movement.
Several weeks ago I started to translate "[Discussion with Eliezer Yudkowsky on AGI interventions](https://www.lesswrong.com/posts/CpvyhFy9WvCNsifkY/dis... | https://www.lesswrong.com/posts/vmXRWoWSeCNX6qPSY/i-currently-translate-agi-related-texts-to-russian-is-that |
# Submit comments on Paxlovid to the FDA (deadline Nov 29th).
1,000 Americans are dying daily from COVID-19. Paxlovid is 89% effective against it. Pfizer submitted an EUA on Paxlovid to the FDA on Nov. 16th. The FDA has yet to schedule a meeting to review this EUA request, 11 days later. Thousands of lives are at stak... | https://www.lesswrong.com/posts/5gzkTjPrrNxCDDARc/submit-comments-on-paxlovid-to-the-fda-deadline-nov-29th |
# [Linkpost] Being Normal by Brian Caplan
[**Being Normal**](https://www.econlib.org/being-normal/)
=========================================================
> *The Principle of Normality: A normal person says what others **say**, but does what others **do**.*
*I found this via a* [*Tweet by Paul Graham*](https://tw... | https://www.lesswrong.com/posts/JtSTBGN8BgJTHqyaa/linkpost-being-normal-by-brian-caplan |
# Why Study Physics?
Physics seems to have a bunch of useful epistemic techniques which haven’t been made very legible yet.
The two big *legible* epistemic techniques in technical fields are Mathematical Proofs, and The Scientific Method. Either derive logically X from some widely-accepted axioms, or hypothesize X an... | https://www.lesswrong.com/posts/qherXDnjKd8upEqhn/why-study-physics |
# Frame Control
*Crossposted from* [*my blog*](https://knowingless.com/)
When I [mention my dad’s abuse](https://knowingless.com/2018/09/21/trauma-narrative/), I mention salient things - physical pain, insults, and controlling behavior. These are “clearly bad” - if I tell you that he often told me I was lazy and woul... | https://www.lesswrong.com/posts/bQ6zpf6buWgP939ov/frame-control |
# Why Bedroom Closets?
Bedrooms, at least in the US, are nearly always constructed with built-in closets, but I don't see why? What's the appeal?
Personally, I don't like them. I want the flexibility to arrange furniture however currently best suits my needs, and a built-in closet permanently reserves a portion of th... | https://www.lesswrong.com/posts/tQjf9qBhPyJFgLWHF/why-bedroom-closets |
# Apparently winning by the bias of your opponents
There's a thing I've sometimes noticed happening in social science debates. In the spirit of [How to Write Quickly While Maintaining Epistemic Rigor](https://www.lesswrong.com/posts/Psr9tnQFuEXiuqGcR/how-to-write-quickly-while-maintaining-epistemic-rigor), let's consi... | https://www.lesswrong.com/posts/RwgNd9vXL5A687dCG/apparently-winning-by-the-bias-of-your-opponents |
# Solve Corrigibility Week
A low-hanging fruit for solving alignment is to dedicate a chunk of time actually trying to solve a sub-problem collectively.
To that end, I’ve broken up researching the sub-problem of corrigibility into two categories in this [google doc](https://docs.google.com/document/d/1vMucI0L3ssECOpJ... | https://www.lesswrong.com/posts/Lv3emECEjkCSHG7L7/solve-corrigibility-week |
# The bonds of family and community: Poverty and cruelty among Russian peasants in the late 19th century
[*Village Life in Late Tsarist Russia*](https://www.amazon.com/Village-Tsarist-Russia-Religion-America-ebook/dp/B07FZX74DG?tag=jasocraw=20) is an ethnographic account of Russian peasants around 1900. The author, Ol... | https://www.lesswrong.com/posts/mhA4vkeaRn9cpxkag/the-bonds-of-family-and-community-poverty-and-cruelty-among |
# The Rationalists of the 1950s (and before) also called themselves “Rationalists”
**TLDR**
* There’s an organization based in London called the [Rationalist Association](https://en.wikipedia.org/wiki/Rationalist_Association). It was founded in 1885. Historically, it focused on publishing books and articles related... | https://www.lesswrong.com/posts/DtcbfwSrcewFubjxp/the-rationalists-of-the-1950s-and-before-also-called |
# Anthropics and the Universal Distribution
(Cross-posted from [Hands and Cities](https://handsandcities.com/2021/11/28/anthropics-and-the-universal-distribution/). Content warning: especially niche topic.)
Some readers of my recent [sequence on anthropics](https://www.lesswrong.com/posts/RnrpkgSY8zW5ArqPf/sia-greate... | https://www.lesswrong.com/posts/Hcc9fopx7sRexYhhi/anthropics-and-the-universal-distribution |
# Redwood Research is hiring for several roles
[Redwood Research](https://www.redwoodresearch.org/) does applied AI alignment research. We’re ... | https://www.lesswrong.com/posts/g6RW76y5QJAKKDX4t/redwood-research-is-hiring-for-several-roles |
# Watching Myself Program
Michael Malis described an interesting technique for improving his software development workflow:
> One incredibly useful exercise I’ve found is to watch myself program. Throughout the week, I have a program running in the background that records my screen. At the end of the week, I’ll watch... | https://www.lesswrong.com/posts/MAM3pdncCWBrkhnxq/watching-myself-program |
# How to measure FLOP/s for Neural Networks empirically?
Experiments and text by Marius Hobbhahn. I would like to thank Jaime Sevilla, Jean-Stanislas Denain, Tamay Besiroglu, Lennart Heim, and Anson Ho for their feedback and support.
### **Summary:**
We measure the utilization rate of a Tesla P100 GPU for training... | https://www.lesswrong.com/posts/jJApGWG95495pYM7C/how-to-measure-flop-s-for-neural-networks-empirically |
# Comments on Allan Dafoe on AI Governance
*Financial status: This is independent research, now supported by a grant.*
*Epistemic status: Views here are almost entirely my own.*
---
There are some think pieces that lay out a bunch of perspectives with which we might think about a thing. This can be either terrible o... | https://www.lesswrong.com/posts/rjvLpRzd8mqDyZmcF/comments-on-allan-dafoe-on-ai-governance |
# Omicron Variant Post #2
It’s now been three days since [Post #1](https://thezvi.substack.com/p/omicron-variant-post-1-were-fed-its). The situation is evolving rapidly, so it’s time to check in. What have we learned since then? How should we update our beliefs and world models? There will inevitably be mistakes as we... | https://www.lesswrong.com/posts/g2niuEiAPCGBPGfA3/omicron-variant-post-2 |
# Snoo App Thoughts
With our [youngest child](https://norawise.com) we got a Snoo automated bassinet, and overall we've been [very happy with it](https://www.jefftk.com/p/to-the-robobassinet-and-progress). On the other hand, the corresponding phone app has several missing features. Since all of this is software, it sh... | https://www.lesswrong.com/posts/whDjRjgMvpLHvat9N/snoo-app-thoughts |
# Soares, Tallinn, and Yudkowsky discuss AGI cognition
This is a collection of follow-up discussions in the wake of Richard Ngo and Eliezer Yudkowsky's [Sep. 5–8](https://www.lesswrong.com/posts/7im8at9PmhbT4JHsW/ngo-and-yudkowsky-on-alignment-difficulty) and [Sep. 14](https://www.lesswrong.com/s/n945eovrA3oDueqtq/p/h... | https://www.lesswrong.com/posts/oKYWbXioKaANATxKY/soares-tallinn-and-yudkowsky-discuss-agi-cognition |
# Visible Thoughts Project and Bounty Announcement
(**Update Jan. 12**: We released an [FAQ](https://docs.google.com/document/d/1sxNOxvcBWw7XC6MSUUpAjO4Rbu3VcUmCy7PJWQ9Vh5o/edit) last month, with more details. Last updated Jan. 7.)
(**Update Jan. 19:** We now have an example of a [successful partial run](https://docs... | https://www.lesswrong.com/posts/zRn6cLtxyNodudzhw/visible-thoughts-project-and-bounty-announcement |
# Seeking Truth Too Hard Can Keep You from Winning
LW-style rationality is about [winning](https://www.lesswrong.com/posts/4ARtkT3EYox3THYjF/rationality-is-systematized-winning). Instrumentally, most rationalists think they can win, in part, by seeking [truth](https://www.lesswrong.com/tag/truth-semantics-and-meaning)... | https://www.lesswrong.com/posts/5HtJR7wjYH6z8HgWL/seeking-truth-too-hard-can-keep-you-from-winning |
# AI Governance Fundamentals - Curriculum and Application
_[Edit: The course has received major updates since this post was made. **See [here](https://course.aisafetyfundamentals.com/governance) for the most recent course information and curriculum**. I've removed the largely outdated information from this post.]_
Wh... | https://www.lesswrong.com/posts/hSc4yMamMzrHfJrKF/ai-governance-fundamentals-curriculum-and-application |
# My take on higher-order game theory
This is how I currently think about [higher-order game theory](https://www.alignmentforum.org/posts/FPML8k4QtjJxk3Y4M/confusions-re-higher-level-game-theory), the study of agents thinking about agents thinking about agents....
This post doesn't add any new big ideas beyond what w... | https://www.lesswrong.com/posts/Cw84NJXAma85AmpnH/my-take-on-higher-order-game-theory |
# Looking for reasoned discussion on Geert Vanden Bossche's ideas? +6 Months on...
I'm again looking for reasoned discussion on Geert Vanden Bossche's ideas. This time we can compare to our discussion 6 months ago.
See: [https://www.lesswrong.com/posts/mjgcFJYFFgjqod9gi/looking-for-reasoned-discussion-on-geert-vande... | https://www.lesswrong.com/posts/YhejgJjP7sRcahSBR/looking-for-reasoned-discussion-on-geert-vanden-bossche-s-1 |
# Tears Must Flow
This Thanksgiving I was with my girlfriend's family. Turkey was served. I held back tears for a little and finished my food. But I couldn't hold them back very long. I ended up going down to the basement to cry. I cried off and on for over an hour. My girlfriend held me. Eventually we went back upsta... | https://www.lesswrong.com/posts/KnSWYTHEFSGqnTgju/tears-must-flow |
# Reneging Prosocially
*Author's note: this essay was originally published elsewhere in 2019. It's now being permanently rehosted at this link. I'll be rehosting a small number of other upper-quintile essays from my past writing over the coming weeks.*
* * *
Quite often, I will make an agreement, and then find mysel... | https://www.lesswrong.com/posts/sjRG35aq5fosJ6mdG/reneging-prosocially |
# Relying on Future Creativity
I used to write songs for a rock band. Sometimes I'd have a song written, thinking it was my best work ever, then when it came time to rehearse, we'd realize it wasn't going to work out. Maybe we didn't have the instruments to do it justice (we were a three piece), or it was out of my co... | https://www.lesswrong.com/posts/kZY4PdauuAFWp5Qkt/relying-on-future-creativity |
# DC Winter Solstice
Solstice is a time to sing songs, tell stories, contemplate how dark and cold the universe is, and remind ourselves that together, we can make it less so.
The DC Solstice will be held **outdoors**, in a backyard at a private residence in Silver Spring. It is not in walking distance of a Metro sto... | https://www.lesswrong.com/events/CNmcKmgmWxqbTNfxS/dc-winter-solstice |
# Infra-Bayesian physicalism: a formal theory of naturalized induction
*This is joint work by Vanessa Kosoy and Alexander "Diffractor" Appel.*
*For the proofs, see [1](https://www.alignmentforum.org/posts/cj3PRu8QoFm4BA8oc/infra-bayesian-physicalism-proofs-part-i) and [2](https://www.alignmentforum.org/posts/CPr8bRGek... | https://www.lesswrong.com/posts/gHgs2e2J5azvGFatb/infra-bayesian-physicalism-a-formal-theory-of-naturalized |
# Experience with Cue Covid Testing
Several months ago my work started offering at-home covid-19 rapid molecular testing via [Cue](https://www.cuehealth.com/products/how-cue-detects-covid-19/). Now that it's possible to [buy](https://shop.cuehealth.com/pages/memberships) these kits as an individual I wanted to write s... | https://www.lesswrong.com/posts/JYCHxkat9Z2PpbtKj/experience-with-cue-covid-testing |
# How do you write original rationalist essays?
I really enjoy Scott Alexander's and Paul Graham's essays. How can I practice to learn to write as they do?
I'm getting pretty okay at writing tutorials, where I just walk people through the process of completing some project. I'm also okay at research-based posts - it'... | https://www.lesswrong.com/posts/Ffm6CfHPDthxozAua/how-do-you-write-original-rationalist-essays |
# FDA Votes on Molunpiravir
While [Paxlovid Remains Illegal](https://thezvi.wordpress.com/2021/11/24/paxlovid-remains-illegal-11-24-update/) and is expected to remain illegal for at least several weeks, the FDA did manage to finally meet to discuss whether or not to legalize the _other_ Covid-19 treatment pill, Merck’... | https://www.lesswrong.com/posts/8Jcyb9d3SReFx8rcS/fda-votes-on-molunpiravir |
# The Limits Of Medicine - Part 1 - Small Molecules
Much of medicine relies on what is dubbed "[small molecules](https://en.wikipedia.org/wiki/Small_molecule)".
"Molecules" since they are atoms tightly bound together, forming what is easily seen as a unitary whole, as opposed to e.g. [lipoproteins](https://en.wikiped... | https://www.lesswrong.com/posts/HiibTXneebtAuNjh4/the-limits-of-medicine-part-1-small-molecules |
# Hypotheses about Finding Knowledge and One-Shot Causal Entanglements
*Epistemic status: my own thoughts I've thought up in my own time. They may be quite or very wrong! I am likely not the first person to come to these ideas. All of my main points here are just hypotheses which I've come to by the reasoning stated b... | https://www.lesswrong.com/posts/YKBbqMXSKetTQ2HBW/hypotheses-about-finding-knowledge-and-one-shot-causal |
# Taking Clones Seriously
**What this is not:** an exhortation to dig up Von Neumann and clone him
**What this is:** an exhortation to think seriously about how to decide whether to dig up Von Neumann and clone him
Having tried to clarify that I am not presently advocating for cloning, and am far more concerned with... | https://www.lesswrong.com/posts/s5ce8BaTAprgDBHni/taking-clones-seriously |
# Open & Welcome Thread December 2021
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 invi... | https://www.lesswrong.com/posts/E8fLwwmeGaX3nqzhK/open-and-welcome-thread-december-2021 |
# Biology-Inspired AGI Timelines: The Trick That Never Works
\- 1988 -
---------
**Hans Moravec:** Behold my book *Mind Children.* Within, I project that, in 2010 or thereabouts, we shall achieve strong AI. I am not calling it "Artificial General Intelligence" because this term will not be coined for another 15 ye... | https://www.lesswrong.com/posts/ax695frGJEzGxFBK4/biology-inspired-agi-timelines-the-trick-that-never-works |
# The 2020 Review
Today is the first day of the LessWrong 2020 Review. At the end of each year, we take a look at the best posts from the *previous* year, and reflect on which of them have stood the test of time.
As we navigate the 21st century, a key issue is that we’re pretty confused about which intellectual work... | https://www.lesswrong.com/posts/M9kDqF2fn3WH44nrv/the-2020-review |
# Common Probability Distributions
When we output a forecast, we're either explicitly or implicitly outputting a _probability distribution_.
For example, if we forecast the AQI in Berkeley tomorrow to be "around" 30, plus or minus 10, we implicitly mean some distribution that has most of its probability mass between ... | https://www.lesswrong.com/posts/jmq3mon8TSC99ittm/common-probability-distributions |
# AXRP Episode 12 - AI Existential Risk with Paul Christiano
[YouTube link](https://www.youtube.com/watch?v=3L4czdIa8Tg&list=PLmjaTS1-AiDeqUuaJjasfrM6fjSszm9pK&index=12)
This podcast is called AXRP, pronounced axe-urp and short for the AI X-risk Research Podcast. Here, I ([Daniel Filan](https://danielfilan.com/)) hav... | https://www.lesswrong.com/posts/krsjmpDB4kgDq6pdu/axrp-episode-12-ai-existential-risk-with-paul-christiano |
# Morality is Scary
I'm worried that many AI alignment researchers and other LWers have a view of how human morality works, that really only applies to a small fraction of all humans (notably moral philosophers and themselves). In this view, people know or at least suspect that they are confused about morality, and ar... | https://www.lesswrong.com/posts/y5jAuKqkShdjMNZab/morality-is-scary |
# Sydney AI Safety Fellowship
I’m excited to announce the launch of the Sydney AI Safety Fellowship which will provide fellows from Australia and New Zealand the opportunity to pursue projects in AI Safety or spend time upskilling. These projects may be technical projects, projects related to policy, or movement build... | https://www.lesswrong.com/posts/jotEekAQxmrwcMf9e/sydney-ai-safety-fellowship |
# Covid Prediction Markets at Polymarket
We now have some useful prediction markets up on Covid issues, so it’s worth looking at what they say and thinking about what other markets we could generate. I encourage you to suggest additional markets in the comments, with as much detail as possible.
[As I wrote a while ag... | https://www.lesswrong.com/posts/M9ngsjEMGjaYs84W8/covid-prediction-markets-at-polymarket |
# Built-in Measuring Spoons
In making the kids breakfast each morning, which has moved on from [crepes](https://www.jefftk.com/p/eggy-crepes) and [waffles](https://www.jefftk.com/p/eggy-waffles) to pancakes, I want a consistent amount of baking powder. I realized that the container has a built-in "measuring spoon". If... | https://www.lesswrong.com/posts/6HnPMTmg79q8mDppC/built-in-measuring-spoons |
# Omicron Post #3
I have split off the Omicron post from the weekly post, which covers other aspects of Covid. I made judgment calls on where to put a few things.
Omicron continues to spread quickly, but the _news surrounding_ Omicron is clearly slowing down, so these updates can hopefully space out a bit more.
Past... | https://www.lesswrong.com/posts/AWRzZTwGxkaqeppJf/omicron-post-3 |
# Covid 12/2: But Aside From That
This weekly post is for things that are *not* Omicron. For the Omicron news, see my Omicron updates ([#1](https://thezvi.wordpress.com/2021/11/26/omicron-variant-post-1-were-fed-its-never-over/), [#2](https://thezvi.wordpress.com/2021/11/29/omicron-variant-post-2/), [#3](https://thezv... | https://www.lesswrong.com/posts/tk4sorNNKEM3sqo4n/covid-12-2-but-aside-from-that |
# "Infohazard" is a predominantly conflict-theoretic concept
Nick Bostrom [writes](https://www.nickbostrom.com/information-hazards.pdf) about "information hazards", or "infohazards":
> Information hazards are risks that arise from the dissemination or the potential dissemination of true information that may cause har... | https://www.lesswrong.com/posts/T6Zpkc2AuJAr8PfJN/infohazard-is-a-predominantly-conflict-theoretic-concept |
# Forecasting Newsletter: November 2021
Highlights
----------
* Polymarket sees [record-high swings](https://twitter.com/Domahhhh/status/1461503351669989386)
* Replication Markets [pays out $142k in forecaster rewards](https://www.replicationmarkets.com/index.php/2021/11/16/resolutions-for-replication-markets/)
*... | https://www.lesswrong.com/posts/zHkBDvnvCA3y6XPJS/forecasting-newsletter-november-2021 |
# Formalizing Policy-Modification Corrigibility
In [*Corrigibility Can Be VNM-Incoherent*](https://www.lesswrong.com/posts/WCX3EwnWAx7eyucqH/corrigibility-can-be-vnm-incoherent), I operationalized an agent's corrigibility$$ as our ability to modify the agent so that it follows different policies. In the summer of 2020... | https://www.lesswrong.com/posts/RAnb2A5vML95rBMyd/formalizing-policy-modification-corrigibility |
# Second-order selection against the immortal
*Cross-posted from* [*Telescopic Turnip*](https://www.telescopic-turnip.net/hogwash/second-order-selection-against-the-immortal/)*.*
In his [recent review](https://astralcodexten.substack.com/p/book-review-lifespan) of *Lifespan*, Scott Alexander writes:
> [Algernon’s La... | https://www.lesswrong.com/posts/xL7yjvajki4eAf5JY/second-order-selection-against-the-immortal |
# Russian x-risks newsletter fall 2021
It is estimated that excess morality during Covid pandemics will reach 1 million in Russia to the end of 2021, according to demograph A.Raksha. He also said that in October 2021 daily morality in Russia was highest since 1945. He uses numbers from local death-registration offices... | https://www.lesswrong.com/posts/dFhumw2Ln683jzuwt/russian-x-risks-newsletter-fall-2021 |
# $100/$50 rewards for good references
*With thanks to Rohin Shah.*
Dear LessWrongers, this is an opportunity to make money and help with AI alignment.
We're looking for specific AI capabilities; has anyone published on the following subject:
* **Generating multiple reward functions or policies from the same set of... | https://www.lesswrong.com/posts/ex2qcux8TQXigGAfv/usd100-usd50-rewards-for-good-references |
# The Learning System
Civilization is kept afloat by a massive, decentralized body of often unseen knowledge.
This dark mass is made up of innumerable pieces of know-how accumulated by people mostly stumbling around, observing each other and the way things work. It's what's lodged in the head of the East German handy... | https://www.lesswrong.com/posts/FhyCMaQfe9sjW6o9f/the-learning-system |
# Browser Engines
While there are many web browsers, they're almost all wrappers around the only three actively developed full engines. Those are:
* [Gecko](https://en.wikipedia.org/wiki/Gecko_(software)) ([source code](https://hg.mozilla.org/mozilla-central/)): Developed by Mozilla for [Firefox](https://en.wikiped... | https://www.lesswrong.com/posts/ve5BZ3Cn9XybzbFRR/browser-engines |
# [Linkpost] A General Language Assistant as a Laboratory for Alignment
This is a linkpost for a recent paper from Anthropic: [A General Language Assistant as a Laboratory for Alignment](https://arxiv.org/abs/2112.00861).
Abstract:
> Given the broad capabilities of large language models, it should be possible to wor... | https://www.lesswrong.com/posts/dktT3BiinsBZLw96h/linkpost-a-general-language-assistant-as-a-laboratory-for |
# Shulman and Yudkowsky on AI progress
This post is a transcript of a discussion between Carl Shulman and Eliezer Yudkowsky, following up on [a conversation with Paul Christiano and Ajeya Cotra](https://www.lesswrong.com/posts/7MCqRnZzvszsxgtJi/christiano-cotra-shulman-and-yudkowsky-on-ai-progress).
Color key:
| Cha... | https://www.lesswrong.com/posts/sCCdCLPN9E3YvdZhj/shulman-and-yudkowsky-on-ai-progress |
# Should we postpone getting a booster due to Omicron, till there are Omicron-specific boosters?
Wai Dai writes over in the comments at [Omicron Variant Post #1: We’re F***ed, It’s Never Over](https://www.lesswrong.com/posts/6apSCHHuWyxK635pE/omicron-variant-post-1-we-re-f-ed-it-s-never-over?commentId=MbvSmohprgXaGCDu... | https://www.lesswrong.com/posts/vA5EmWEjMqwR53Pwn/should-we-postpone-getting-a-booster-due-to-omicron-till |
# Lars Doucet's Georgism series on Astral Codex Ten
*\[Lars Doucet won this year’s* [*Book Review Contest*](https://astralcodexten.substack.com/p/book-review-contest-winners) *with his review of Henry George’s* [*Progress and Poverty*](https://astralcodexten.substack.com/p/your-book-review-progress-and-poverty)*. Sinc... | https://www.lesswrong.com/posts/XtRAkvvaQSaQEyASj/lars-doucet-s-georgism-series-on-astral-codex-ten |
# Misc. questions about EfficientZero
Perhaps these can be thought of as homework questions -- when I imagine us successfully making AI go well, I imagine us building expertise such that we can answer these questions quickly and easily. Before I read the answers I'm going to think for 10min or so about each one and po... | https://www.lesswrong.com/posts/rtBpBNgXjwtsLJDbG/misc-questions-about-efficientzero-1 |
# Agency: What it is and why it matters
_\[ETA: I'm deprioritizing completing this sequence because it seems that other people are writing good similar stuff. In particular, see e.g. [https://www.lesswrong.com/posts/kpPnReyBC54KESiSn/optimality-is-the-tiger-and-agents-are-its-teeth](https://www.lesswrong.com/posts/kpP... | https://www.lesswrong.com/posts/qJBkcGW4GitfQ4BBy/agency-what-it-is-and-why-it-matters |
# Agents as P₂B Chain Reactions
***tl;dr:** Sometimes planners successfully* [*P~2~B*](https://www.lesswrong.com/posts/CAwwFpbteYBQw2Gkp/p-b-plan-to-p-b-better)*, kicking off a self-sustaining chain reaction / feedback loop of better and better plans (made possible by better and better world-models, more and more reso... | https://www.lesswrong.com/posts/oiftkZnFBqyHGALwv/agents-as-p-b-chain-reactions |
# A Generalization of ROC AUC for Binary Classifiers
Suppose you have a binary classifier. It looks at things and tries to guess whether they’re Dogs or Not Dogs.
More precisely, the classifier outputs a numeric *score*, which is higher for things it thinks are more likely to be Dogs.
There are a bunch of ways to as... | https://www.lesswrong.com/posts/7DWn54otFEBYyDQFi/a-generalization-of-roc-auc-for-binary-classifiers |
# Covid Christmas
Every year, my extended family gets together for Christmas. Several days of lots of people in the same place, eating, playing games, making music, and generally spending time together. We skipped last year, because of covid, and are thinking about what makes sense to do this year.
If rates had staye... | https://www.lesswrong.com/posts/8J7md3L2vuJoWfF6g/covid-christmas |
# Privacy and Manipulation
Previously:
* [“Can you keep that confidential? How do you know?”](https://www.lesswrong.com/posts/rz73eva3jv267Hy7B/can-you-keep-this-confidential-how-do-you-know)
* [Parameters of Privacy](https://www.lesswrong.com/posts/afqz78RTjGB9a7MPe/parameters-of-privacy#i__Am_I_making_a_promise... | https://www.lesswrong.com/posts/YGB5f7kioXDXNAwLx/privacy-and-manipulation-1 |
# Behavior Cloning is Miscalibrated
Behavior cloning (BC) is, put simply, when you have a bunch of human expert demonstrations and you train your policy to maximize likelihood over the human expert demonstrations. It’s the simplest possible approach under the broader umbrella of Imitation Learning, which also includes... | https://www.lesswrong.com/posts/BgoKdAzogxmgkuuAt/behavior-cloning-is-miscalibrated |
# What have your romantic experiences with non-EAs/non-Rationalists been like?
Epistemic status: trawlin' for anecdotes/commiserations; satiating-morbid-curiosity
I'm also interested in non-romantic relationships, but I'm *particularly* interested in hearing about long term relationships between folks within EA and/o... | https://www.lesswrong.com/posts/kijoFAKFGKr6Tq29z/what-have-your-romantic-experiences-with-non-eas-non |
# Scott Alexander's "Ivermectin: Much More Than You Wanted To Know"
This is a linkpost, created for the 2021 Review.
> I know I’m two months late here. Everyone’s already made up their mind and moved on to other things.
>
> But here’s my pitch: this is one of the most carefully-pored-over scientific issues of our ti... | https://www.lesswrong.com/posts/3iMXHJCWNbmxr9peT/scott-alexander-s-ivermectin-much-more-than-you-wanted-to |
# Interpreting Yudkowsky on Deep vs Shallow Knowledge
*Here is an exploration of what Eliezer Yudkowsky means when he writes about deep vs shallow patterns (although I’ll be using "knowledge" instead of "pattern" for reasons explained in the next section). Not about any specific pattern Yudkowsky is discussing, mind y... | https://www.lesswrong.com/posts/GSBCw94DsxLgDat6r/interpreting-yudkowsky-on-deep-vs-shallow-knowledge |
# Explicit model visualizations
What is this about?
===================
I believe explicit model visualizations explain systems better, and if we had the tools to effectively work with them, they could improve the quality of debates and increase rationality in institutions.
In [Lies, Damn Lies and Fabricated Options... | https://www.lesswrong.com/posts/HhZDfDDcFmmtKqp6f/explicit-model-visualizations |
# Are limited-horizon agents a good heuristic for the off-switch problem?
(This is my first post, sorry if this is covered elsewhere.)
Implicit in the problem of a superhuman AI's reward being misspecified and turning everything into paperclips is the fact that the agent is optimizing over e.g. "number of paperclips"... | https://www.lesswrong.com/posts/itTLCFj5NCHhFbK2Q/are-limited-horizon-agents-a-good-heuristic-for-the-off |
# ML Alignment Theory Program under Evan Hubinger
In the past six weeks, the Stanford Existential Risks Initiative (SERI) has been running a trial for the “ML Alignment Theory Scholars” (MATS) program. Our goal is to increase the number of people working on alignment theory, and to do this, we’re running a scholars pr... | https://www.lesswrong.com/posts/FpokmCnbP3CEZ5h4t/ml-alignment-theory-program-under-evan-hubinger |
# Anti-correlated causation
### Example 1: exercise correlates with weight gain?
Let's assume that, when holding diet constant, exercise causes weight loss in the general population. Basic military training causes soldiers to exercise, but it also [may cause weight gain](https://militaryhealth.bmj.com/content/early/2... | https://www.lesswrong.com/posts/ZAxzcCrKdDAQYKnii/anti-correlated-causation |
# A Framework to Explain Bayesian Models
[Bayesian Networks](https://en.wikipedia.org/wiki/Bayesian_network#:~:text=A%20Bayesian%20network%20(also%20known,directed%20acyclic%20graph%20(DAG).) are used to represent uncertainty and probabilistic relations between variables. They have an appealing graphical representatio... | https://www.lesswrong.com/posts/SPfZiEwHotPncJBLz/a-framework-to-explain-bayesian-models |
# Modeling Failure Modes of High-Level Machine Intelligence
*This post, which deals with some widely discussed failure modes of transformative AI, is part 7 in our* [*sequence on Modeling Transformative AI Risk*](https://www.alignmentforum.org/s/aERZoriyHfCqvWkzg)*. In this series of posts, we are presenting a prelimi... | https://www.lesswrong.com/posts/3Eq5Rq5uQ97kt8B8f/modeling-failure-modes-of-high-level-machine-intelligence |
# Life, struggle, and the psychological fallout from COVID
I have a friend in the effective altruism community who is facing an unusually slippery and disheartening struggle at the moment. This person is a long-time contributor to the community, and is held up as exemplary by other community members. Due to their inte... | https://www.lesswrong.com/posts/xZ6Cnp9kx4NP4LHYP/life-struggle-and-the-psychological-fallout-from-covid |
# Omicron Post #4
Previous Omicron updates: [#1](https://thezvi.wordpress.com/2021/11/26/omicron-variant-post-1-were-fed-its-never-over/), [#2](https://thezvi.wordpress.com/2021/11/29/omicron-variant-post-2/), [#3](https://thezvi.wordpress.com/2021/12/02/omicron-post-3/). [Last weekly non-Omicron update](https://thezv... | https://www.lesswrong.com/posts/gtYGFZf6ojaDfpduD/omicron-post-4 |
# A Possible Resolution To Spurious Counterfactuals
Spurious counterfactuals (perhaps an easier handle than "the lobian inference issues described in Section 2.1 [here, in the Embedded Agency paper](https://arxiv.org/pdf/1902.09469.pdf) ), are important to address because they lead to inference problems. Having a func... | https://www.lesswrong.com/posts/TnkDtTAqCGetvLsgr/a-possible-resolution-to-spurious-counterfactuals |
# Are there alternative to solving value transfer and extrapolation?
*Thanks to Rebecca Gorman for the discussions that inspired these ideas.*
We recently argued that full understanding of value extrapolation[^ve] was [necessary and almost sufficient](https://www.lesswrong.com/posts/3e6pmovj6EJ729M2i/general-alignmen... | https://www.lesswrong.com/posts/DjTKMEwRqpuKkJzTo/are-there-alternative-to-solving-value-transfer-and |
# More Christiano, Cotra, and Yudkowsky on AI progress
This post is a transcript of a discussion between Paul Christiano, Ajeya Cotra, and Eliezer Yudkowsky (with some comments from Rob Bensinger, Richard Ngo, and Carl Shulman), continuing from [1](https://www.lesswrong.com/posts/vwLxd6hhFvPbvKmBH/yudkowsky-and-christ... | https://www.lesswrong.com/posts/fS7Zdj2e2xMqE6qja/more-christiano-cotra-and-yudkowsky-on-ai-progress |
# Declustering, reclustering, and filling in thingspace
Eliezer has [talked about thingspace](https://www.lesswrong.com/posts/WBw8dDkAWohFjWQSk/the-cluster-structure-of-thingspace) - an abstract space in which objects lie, defined by their properties on some scales. In thingspace, there are some more-or-less natural c... | https://www.lesswrong.com/posts/WikzbCsFjpLTRQmXn/declustering-reclustering-and-filling-in-thingspace |
# Leaving Orbit
Duncan says:
> At CFAR workshops, we often installed this as an explicit local norm: "If at any point you want to convey 'hey, I'm bowing out of this conversation now, but it's not intended to be any sort of reflection on you or the topic, I'm not making a statement, I'm just doing what's good for me ... | https://www.lesswrong.com/posts/Bgt236xrd5j2uAnQJ/leaving-orbit |
# Dear Self; We Need To Talk About Social Media
[Last year](https://acesounderglass.com/2020/05/31/turns-out-interruptions-are-bad-who-knew/) I discovered, much to my chagrin, that always-on internet socializing was costly for me. This was inconvenient both because I’d spent rather a lot of time singing the praises of... | https://www.lesswrong.com/posts/fwNskn4dosKng9BCB/dear-self-we-need-to-talk-about-social-media |
# Retail Investor Advantages
There are some advantages to being a retail investor that might make it possible for you to beat the market without violating EMH.
Information Acquisition Cost
----------------------------
If you are an expert in some domain, it costs less for you to assess information in that domain tha... | https://www.lesswrong.com/posts/TdzxoS4AMgwysnixt/retail-investor-advantages |
# Considerations on interaction between AI and expected value of the future
Some thoughts about the ‘default’ trajectory of civilisation and how AI will affect the likelihood of different outcomes.
### Is the default non-extinction outcome utopic or dystopic?
**Arguments for dystopia:**
1. The world seems dystopi... | https://www.lesswrong.com/posts/Dr3owdPqEAFK4pq8S/considerations-on-interaction-between-ai-and-expected-value |
# Interviews on Improving the AI Safety Pipeline
I conducted the following interviews to better understand how we might aim to improve the AI safety pipeline. It is very hard to summarise these interviews as everyone has different ideas of what needs to be done. However, I suppose there is some value in knowing this i... | https://www.lesswrong.com/posts/Q3c8bJNrfF3hEb9Kx/interviews-on-improving-the-ai-safety-pipeline |
# Counting Lightning
When Samir returned to the base JiuJang waited by the entrance. There was no point in denying he had disobeyed JiuJang's warning, so he was upfront about it.
"I went outside," Samir admitted. "It was raining heavily. The wind blew so strong I barely could hear anything. And every second or so, li... | https://www.lesswrong.com/posts/cLBDeEnMsf4vfxvzc/counting-lightning |
# Randomness in Science
I.
Humans have a randomness problem. We are bad at generating randomness as individuals (try to out-random a computer in [Man vs. Machine](http://www.loper-os.org/bad-at-entropy/manmach.html)) and in the aggregate - ask a group of people to choose a “random” number between 1-20 and the most c... | https://www.lesswrong.com/posts/WTgHBsvrLRBBB5Lid/randomness-in-science |
# Let's buy out Cyc, for use in AGI interpretability systems?
(UPDATE JANUARY 2026: This post is terribly outdated in several respects; see [Intro to Brain-Like-AGI Safety §15.2.2.1](https://www.lesswrong.com/posts/tj8AC3vhTnBywdZoA/intro-to-brain-like-agi-safety-15-conclusion-open-problems-1#15_2_2_1_The__Make_the_bi... | https://www.lesswrong.com/posts/nqFS7h8BE6ucTtpoL/let-s-buy-out-cyc-for-use-in-agi-interpretability-systems |
# Theoretical Neuroscience For Alignment Theory
*This post was written under Evan Hubinger’s direct guidance and mentorship, as a part of the* [*Stanford Existential Risks Institute ML Alignment Theory Scholars (MATS) program*](https://www.alignmentforum.org/posts/FpokmCnbP3CEZ5h4t/ml-alignment-theory-program-under-ev... | https://www.lesswrong.com/posts/ZJY3eotLdfBPCLP3z/theoretical-neuroscience-for-alignment-theory |
# What makes for a good "argument"? (Request for thoughts and comments)
With colleagues at CMU, we've been looking at the ways in which people make arguments. The goal here is to look at what we call "argument making in the wild", i.e., to try to build a taxonomy of the different ways people make arguments, whether or... | https://www.lesswrong.com/posts/SaHg7otE29ncbiXFC/what-makes-for-a-good-argument-request-for-thoughts-and |
# Interpreting the Biobot Spike
Boston has a really interesting program where it [measures covid RNA frequency during sewage treatment](https://www.mwra.com/biobot/biobotdata.htm). It's been running since the beginning of the pandemic:
[](https://www.jefftk.co... | https://www.lesswrong.com/posts/yaC7Eexvm6dzSiSrK/interpreting-the-biobot-spike |
# [AN #170]: Analyzing the argument for risk from power-seeking AI
Listen to this newsletter on **[The Alignment Newsletter Podcast](http://alignment-newsletter.libsyn.com/)**.
Alignment Newsletter is a weekly publication with recent content relevant to AI alignment around the world. Find all Alignment Newsletter **[... | https://www.lesswrong.com/posts/PC6QavgNDQjHbutAq/an-170-analyzing-the-argument-for-risk-from-power-seeking-ai |
# Finding the multiple ground truths of CoinRun and image classification
## Research projects
I'm planning to start two research projects on [model splintering/reward generalisation](https://www.lesswrong.com/posts/k54rgSg7GcjtXnMHX/model-splintering-moving-from-one-imperfect-model-to-another-1) and [learning the pre... | https://www.lesswrong.com/posts/oCWk8QpjgyqbFHKtK/finding-the-multiple-ground-truths-of-coinrun-and-image |
# COVID and the holidays
6 months ago I wrote about how 30-year-olds should basically [go back to normal](https://www.lesswrong.com/posts/rWXNvq96jTH9YAm7J/let-s-go-back-to-normal) and no longer take many COVID precautions.
The holidays make this policy somewhat worse, because you can infect your family members. Furt... | https://www.lesswrong.com/posts/GzzJZmqxcqg5KFf8r/covid-and-the-holidays |
# Introduction to inaccessible information
*This post was written under Evan Hubinger's direct guidance and mentorship, as a part of the [Stanford Existential Risks Institute ML Alignment Theory Scholars (MATS) program](https://www.lesswrong.com/posts/FpokmCnbP3CEZ5h4t/ml-alignment-theory-program-under-evan-hubinger).... | https://www.lesswrong.com/posts/CYKeDjD7FEvAnzBBF/introduction-to-inaccessible-information |
# Stop arbitrarily limiting yourself
*Crossposted from* [*my blog*](https://unoptimal.com/)
My roommate types with one finger; he literally only uses his right index finger. I made him do a typing test and his WPM turned out to be around 30-40, which is honestly higher than I thought it’d be.
When I first noticed th... | https://www.lesswrong.com/posts/FKQuWBEB5dap3gHG2/stop-arbitrarily-limiting-yourself |
# Supervised learning and self-modeling: What's "superhuman?"
**I**
We don't just want aligned AI to do what a human would do. We've got plenty of humans lying around all over the place, if that's all we need. The fancy promise of future AI is that it will deliver *superhuman* performance.
Sometimes the notion of "... | https://www.lesswrong.com/posts/pz84sQKsgg3GBHQpd/supervised-learning-and-self-modeling-what-s-superhuman |
# [MLSN #2]: Adversarial Training
As part of a larger community building effort, I am writing a safety newsletter which is designed to cover empirical safety research and be palatable to the broader machine learning research community. You can [subscribe here](https://newsletter.mlsafety.org/) or follow the newsletter... | https://www.lesswrong.com/posts/7GQZyooNi5nqgoyyJ/mlsn-2-adversarial-training |
# The end of Victorian culture, part I: structural forces
If a Victorian were transported to 2021, he would be amazed by the progress of technology. He’d be fascinated by the triumphs and challenges of democratic politics. But he would probably also say that our culture had been taken over by the Devil.
For good reas... | https://www.lesswrong.com/posts/xkShF3BQBnhpDMgGS/the-end-of-victorian-culture-part-i-structural-forces |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.