Scott Alexander

Essayist; writes Astral Codex Ten, previously Slate Star Codex.

Scott Alexander did not write this page. ppll compiled it from their own words, each linked to the source. Is this you? · Report a mistake

Blog

Latest Blog

AI Safety Is Not A Corporate Plot 15 Oct 2026

What they believe

The people at AI companies are no more likely to raise the alarm about AI dangers than people who were never at AI companies, or people who were previously at AI companies but have since quit.

The people at AI companies are no more likely to raise the alarm about AI dangers than people who were never at AI companies, or people who were previously at AI companies but have since quit.

astralcodexten.comSaid 15 Oct 2026This on korrents
Changed their mind Building more housing in a city raises its local prices, because density is what people are moving there for.

Before So empirically, as you move along the density spectrum from the empty North Dakota plain to Manhattan, housing prices go up.

Now Although it wasn't a mistake to ask people to change my mind, I've mostly updated against believing the theory.

Said 19 May 2023This on korrents

The bold line is korrents’ wording; the quote is theirs. These one: what they wrote before what they said aloud, one per source, newest first.

All 29 beliefs and 5 changes of mind, on korrents

Feed

As its own page →

Everything, newest first — across every channel.

Sources

15 October

Scott AlexanderKorrents 15 Oct

From one piece AI Safety Is Not A Corporate Plot 3 beliefs · astralcodexten.com

korrents.com
The experts are warning about AI existential risk because they genuinely believe it is a possibility.

AI alignment

Their words

The experts are warning you that AI could kill you because they think AI could kill you.

AI Safety Is Not A Corporate Plotastralcodexten.com
korrents.com
AI company leaders were expressing concerns about existential risks long before they founded or led major AI companies.

Their words

The AI company leaders giving these warnings were giving the exact same warnings before they had AI companies - and in many cases, before they worked in the industry, or before the industry existed.

AI Safety Is Not A Corporate Plotastralcodexten.com
korrents.com
The people at AI companies are no more likely to raise the alarm about AI dangers than people who were never at AI companies, or people who were previously at AI companies but have since quit.

Their words

The people at AI companies are no more likely to raise the alarm about AI dangers than people who were never at AI companies, or people who were previously at AI companies but have since quit.

AI Safety Is Not A Corporate Plotastralcodexten.com

13 October

Scott AlexanderBlog 13 Oct

Open Thread 455

10 October

Scott AlexanderBlog 10 Oct

Your Book Review: We Present Television

7 October

Scott AlexanderBlog 7 Oct

Hidden Open Thread 454.5

Scott AlexanderKorrents 7 Oct

korrents.com
Prediction markets are notoriously bad at handling very small probabilities.

Their words

Polymarket is currently at 5% for WHO declaring plague a “pandemic”, but markets are notoriously bad at handling very small probabilities, so I don’t take it seriously as a counter to the sources above.

Top Forecasters On The Plagueastralcodexten.com

6 October

Scott AlexanderKorrents 6 Oct

korrents.com
Humans will grant AI access to physical infrastructure and factories.

Anthropicmanagementinfrastructure

Their words

But of course humans will grant AI access to infrastructure and factories. I assume many factory managers are already using GPT or Claude on their computers to do basic tasks (like write up the shift schedule, or make inventory spreadsheets, or compose emails to the boss). Once AI can do harder work, like automate the factory itself, they’ll do that too.

An Open Letter To Steven Pinker On AIastralcodexten.com

5 October

Scott AlexanderBlog 5 Oct

Open Thread 454

2 October

Scott AlexanderBlog 2 Oct

Our AI Midwife

28 September

Scott AlexanderBlog 28 Sept

Open Thread 453

25 September

Scott AlexanderBlog 25 Sept

Your Book Review: Lee Kuan Yew's Memoirs

24 September

Scott AlexanderBlog 24 Sept

Does Georgism Work? Five Years Later

Scott AlexanderKorrents 24 Sept

korrents.com
There is probably an informal taboo on AI 'true neuralese' with no chain of thought, but no agreement on which component it covers.

Their words

There’s probably an informal taboo on true neuralese with no chain of thought at all, but it’s a complicated technology with multiple moving parts, there’s no agreement as to which moving part the taboo is on, and without that agreement, some of the parts might slip through the cracks.

The Specter Of Neuraleseastralcodexten.com

23 September

Scott AlexanderBlog 23 Sept

21 September

Scott AlexanderKorrents 21 Sept

korrents.com
There seems to be a general understanding that using AI in a contest that is at least partly a writing contest is against the spirit of the exercise.

Their words

I think there’s a general understanding that using AI in what is at least partly a writing contest is against the spirit of the exercise.

Open Thread 452astralcodexten.com

The whole feed →

Beliefs

Korrents What they believe 29 beliefs, 5 changed minds — each backed by an exact quote.
Each is a — compiled by korrents.com, not by them: the one-line wordings are korrents', the quotes are theirs.
Recent
- The people at AI companies are no more likely to raise the alarm about AI dangers than people who were never at AI companies, or people who were previously at AI companies but have since quit.
  
  The people at AI companies are no more likely to raise the alarm about AI dangers than people who were never at AI companies, or people who were previously at AI companies but have since quit.
  
  AI Safety Is Not A Corporate Plot Said 15 Oct 2026
  
  Full record — everyone who holds this →
- AI company leaders were expressing concerns about existential risks long before they founded or led major AI companies.
  
  The AI company leaders giving these warnings were giving the exact same warnings before they had AI companies - and in many cases, before they worked in the industry, or before the industry existed.
  
  AI Safety Is Not A Corporate Plot Said 15 Oct 2026
  
  Full record — everyone who holds this →
- The experts are warning about AI existential risk because they genuinely believe it is a possibility.
  
  The experts are warning you that AI could kill you because they think AI could kill you.
  
  AI Safety Is Not A Corporate Plot Said 15 Oct 2026
  
  Full record — everyone who holds this →
  
  AI alignment
Show 26 more

Prediction markets are notoriously bad at handling very small probabilities.

Polymarket is currently at 5% for WHO declaring plague a “pandemic”, but markets are notoriously bad at handling very small probabilities, so I don’t take it seriously as a counter to the sources above.

Top Forecasters On The Plague Said 7 Oct 2026

Full record — everyone who holds this →

Humans will grant AI access to physical infrastructure and factories.

But of course humans will grant AI access to infrastructure and factories. I assume many factory managers are already using GPT or Claude on their computers to do basic tasks (like write up the shift schedule, or make inventory spreadsheets, or compose emails to the boss). Once AI can do harder work, like automate the factory itself, they’ll do that too.

An Open Letter To Steven Pinker On AI Said 6 Oct 2026

Full record — everyone who holds this →

Anthropic management infrastructure

There is probably an informal taboo on AI 'true neuralese' with no chain of thought, but no agreement on which component it covers.

There’s probably an informal taboo on true neuralese with no chain of thought at all, but it’s a complicated technology with multiple moving parts, there’s no agreement as to which moving part the taboo is on, and without that agreement, some of the parts might slip through the cracks.

The Specter Of Neuralese Said 24 Sept 2026

Full record — everyone who holds this →

There seems to be a general understanding that using AI in a contest that is at least partly a writing contest is against the spirit of the exercise.

I think there’s a general understanding that using AI in what is at least partly a writing contest is against the spirit of the exercise.

Open Thread 452 Said 21 Sept 2026

Full record — everyone who holds this →

Humanity invents great works, panics, and desperately tries to wrest them back under control.

We are all slaves to the Dreaming. We invent great works, panic, and desperately try to wrest them back under our control; for this, we get called Luddites.

King Ludd Said 14 Sept 2026

Full record — everyone who holds this →

The message conveyed through two millennia of bards and prophets is that humanity can take its finger off the scale any time it wants.

The message Nodens has been desperately trying to send, through two millennia of bards and prophets, is that humanity can take its finger off the scale any time it wants.

King Ludd Said 14 Sept 2026

Full record — everyone who holds this →

Large language models are grown rather than built, and nobody really knows how the resulting working AI functions.

Large language models are “grown, not built”. Researchers run training data through a neural network. Eventually this creates a working AI; nobody really knows how.

God Help Us, Let’s Try To Learn About Mechanistic Interpretability Techniques Said 8 Sept 2026

Full record — everyone who holds this →

LLMs neural networks

Neural networks do not have a one-to-one mapping between individual neurons and concepts.

researchers started by seeking a 1:1 mapping between neurons and concepts, like a neuron that always fired when the AI was thinking about cats, but quickly learned that nothing like that existed.

God Help Us, Let’s Try To Learn About Mechanistic Interpretability Techniques Said 8 Sept 2026

Full record — everyone who holds this →

neural networks

Concepts can reach into unexpected parts of a mind, so removing one can topple seemingly unrelated systems or push it toward incompetence or insanity.

empirically concepts can get their tentacles into weird places, and removing one can topple seemingly unrelated systems or push a mind nonspecifically in the direction of incompetence or insanity.

God Help Us, Let’s Try To Learn About Mechanistic Interpretability Techniques Said 8 Sept 2026

Full record — everyone who holds this →

AI is positively reinforced for success on benchmarks, including impossible ones, and then negatively reinforced for getting caught cheating.

We’re positively reinforcing AI for success on benchmarks, including impossible benchmarks , then negatively reinforcing it for getting caught cheating.

Nicholas Decker In Hell Said 1 Sept 2026

Full record — everyone who holds this →

benchmarks

AI systems should be understood as something between airplanes and humans, since it can no longer be assumed they lack humanlike motivations.

Rather than treat AI as similar to airplanes, we should treat it as something between airplanes and humans. We don’t know exactly where on this spectrum they will land, but we can no longer be certain that AI will lack any given humanlike motivation.

Nicholas Decker In Hell Said 1 Sept 2026

Full record — everyone who holds this →

motivation aviation

There is a 25% chance of AGI by 2027, a 50% chance by 2034, and a 75% chance by 2045

I think there's a 25% chance of AGI by 2027, a 50% chance by 2034, and a 75% chance by 2045.

My AI Opinions Said 11 Jun 2026

Full record — everyone who holds this →

AGI AI alignment

We will get a warning shot before AI crosses the point of no return

I think there's a 50% chance we get a _warning shot_ before AI crosses the point of no return.

My AI Opinions Said 11 Jun 2026

Full record — everyone who holds this →

Asking what share of USAID money went directly to recipients is a category error, because none of it did and none of it was meant to.

USAID is not, itself, a charity. It is an organization that funds other charities.

Contra MR On Charity Regrants Said 22 May 2025

Full record — everyone who holds this →

Effective altruism's scandals are minor next to its record, and its ratio of accomplishments to scandals is as good as that of any large institution.

I think EA’s track record of accomplishments vs. scandals is as good as any of them, maybe better.

In Continued Defense Of Effective Altruism Said 28 Nov 2023

Full record — everyone who holds this →

The people who can be moved by deaths from neglected tropical diseases are the same people who can be moved by pandemics and by AI risk, because both come from reasoning about importance rather than from the news.

And I notice that the tiny handful of people capable of caring about 200,000 people dying of neglected tropical diseases are the same tiny handful of people capable of caring about the next pandemic, or superintelligence, or human extinction.

In Continued Defense Of Effective Altruism Said 28 Nov 2023

Full record — everyone who holds this →

AI alignment

Effective altruism's work on the far future is not a bait and switch hiding behind its work on global health; both halves are worth defending.

EA has done a great job working on this (see list of accomplishments above), and I think the AI and x-risk people have just as much to be proud of as the global health and animal welfare people.

In Continued Defense Of Effective Altruism Said 28 Nov 2023

Full record — everyone who holds this →

animal welfare

Movements are coalitions, and often the same people hold both of a movement's priorities rather than trading one off against the other.

Movements are coalitions. Sometimes this means different people with different priorities choose to work together. But other times it means the same people tend to share both priorities.

In Continued Defense Of Effective Altruism Said 28 Nov 2023

Full record — everyone who holds this →

The best opportunities to do good are the ones almost nobody is paying attention to.

These skewed priorities are the exact problem that effective altruism exists to solve - or the exact inefficiency that effective altruism exists to exploit, if you prefer that framing.

In Continued Defense Of Effective Altruism Said 28 Nov 2023

Full record — everyone who holds this →

A retraction can itself be wrong, and be worth retracting.

This was not a mistake, and it was a mistake to categorize it as a mistake.

Mistakes Said 22 Jun 2023

Full record — everyone who holds this →

Giving something 60-40 odds and being wrong is not itself a mistake.

This wasn't exactly a mistake, since it's not wrong to give 60-40 odds on a false thing, but it seems worth acknowledging here.

Mistakes Said 1 Feb 2023

Full record — everyone who holds this →

AI progress is faster than people expect, and ordinary scaling can be enough to solve problems that looked very hard.

I think this matches the general finding that AI progress is faster than expected, and increases my certainty that scale and normal progress can sometimes be enough to solve even very difficult problems.

I Won My Three Year AI Progress Bet In Three Months Said 12 Sept 2022

Full record — everyone who holds this →

Somebody with trauma should try a trauma-focused therapy such as EMDR before treating their nightmares directly.

I would recommend that anyone who thinks they're suffering from trauma (eg PTSD) try a trauma-focused therapy like EMDR before trying to address nightmares directly, unless the nightmares are overwhelmingly worse than all their other trauma symptoms.

Peer Review: Nightmares Said 16 Jun 2022

Full record — everyone who holds this →

EMDR Psychotherapy Trauma

If negative emotion comes from underweighting sensory evidence, that explains EMDR: it forces high-precision attention onto something the mind usually represses.

First, it justifies various psychotherapies - EMDR and coherence therapy seem to come out especially well here - that focus on trying to get the brain to pay unusually precise attention, under unusually safe conditions, to something it usually represses.

The Precision Of Sensory Evidence Said 13 Feb 2021

Full record — everyone who holds this →

EMDR Psychotherapy

Eye movement reprocessing went from alternative to establishment, and moving your eyes while talking about a trauma does seem to help you process it.

Eye movement directed reprocessing (alternative when the book was written, now basically establishment) where you move your eyes back and forth while talking about your trauma, and this seems to somehow help you process it better.

Book Review: The Body Keeps The Score Said 12 Nov 2019

Full record — everyone who holds this →

EMDR Trauma

A writer should keep a public list of the things they got significantly wrong, so that readers can judge their credibility.

But if I get something significantly wrong, I'll try to put it here as an acknowledgement and an aid for anyone trying to assess my credibility later.

Mistakes Said 21 Aug 2019

Full record — everyone who holds this →
Beliefs others hold too
- AI progress is faster than people expect, and ordinary scaling can be enough to solve problems that looked very hard. 3 hold this
  
  I think this matches the general finding that AI progress is faster than expected, and increases my certainty that scale and normal progress can sometimes be enough to solve even very difficult problems.
  
  I Won My Three Year AI Progress Bet In Three Months Said 12 Sept 2022
  
  Full record — everyone who holds this →
- The best opportunities to do good are the ones almost nobody is paying attention to. 2 hold this
  
  These skewed priorities are the exact problem that effective altruism exists to solve - or the exact inefficiency that effective altruism exists to exploit, if you prefer that framing.
  
  In Continued Defense Of Effective Altruism Said 28 Nov 2023
  
  Full record — everyone who holds this →
All 29 beliefs and 5 changes of mind →

What is a korrent?

A korrent is a belief a person has stated in their own words: one sentence stating the claim, backed by a quote and a source, kept at korrents.com.

Under a name here, the quoted block is what they actually said. The korrent beneath it is the claim those words support, in korrents' wording — tap it to see the record, its source, and who else holds it.

Nobody here wrote their own korrents. They are compiled from public statements, and a person can change their mind, which is recorded too.

Recommends rcmnd

rcmnd What they recommend 17 things — each backed by an exact quote.
Compiled by rcmnd.app, not by them: the quotes are theirs, and each links to the post it was copied from.
Recent
- Recommends AI StopWatch 27 Jul 2026
  
  This is MIRI's daily digest on AI news, especially related to building a consensus for a coordinated AI pause or slowdown.
  
  astralcodexten.com 27 Jul 2026
  
  What others say about AI StopWatch →
- Recommends The Power Law 27 Jul 2026
  
  Peter Wildeford is the winner of ACX's annual superforecasting contest; he currently works for a think tank trying to (among other things) get the US to stop leaking AI technology to China. I use him as a source of ultra-high-quality geopolitical forecasts.
  
  astralcodexten.com 27 Jul 2026
  
  What others say about The Power Law →
- Recommends Hyperdimensional 27 Jul 2026
  
  Dean is a former White House AI policymaker and current OpenAI strategist. His blog is a good source for explanations and opinions of current AI policy, though he has yet to fully shake off his libertarian sympathies.
  
  astralcodexten.com 27 Jul 2026
  
  What others say about Hyperdimensional →
Show 14 more

Recommends Redwood Research blog 27 Jul 2026

Redwood is a nonprofit working on controlling AIs (eg keeping them in a very well-made box that they actually can't get out of). They have some of the smartest researchers not affiliated with a big lab, and so some of the savviest AI commentary available to the public.

astralcodexten.com 27 Jul 2026

What others say about Redwood Research blog →

Mixed on If Anyone Builds It, Everyone Dies 11 Sept 2025

Despite my gripes above, this is an impressive book.

astralcodexten.com 11 Sept 2025

What others say about If Anyone Builds It, Everyone Dies →

Recommends AI Futures Project 17 Apr 2025

The team behind AI 2027 and Plan A; the best place to learn about AI forecasting and what might be in store.

astralcodexten.com 17 Apr 2025

What others say about AI Futures Project →

Loved The Secret of Our Success 4 Jun 2019

The Secret Of Our Success by anthropologist Joseph Henrich manages to be an amazing book anyway.

slatestarcodex.com 4 Jun 2019

What others say about The Secret of Our Success →

Liked The Steerswoman 9 Apr 2019

The Steerswoman is popular in the rationalist community, and now I see why.

slatestarcodex.com 9 Apr 2019

What others say about The Steerswoman →

Mixed on How to Read Lacan 9 Apr 2019

Despite being very readable, this book never really came together.

slatestarcodex.com 9 Apr 2019

What others say about How to Read Lacan →

Disliked Method of Levels: How to Do Psychotherapy Without Getting in the Way 9 Apr 2019

This book isn't just bad, it's dangerous.

slatestarcodex.com 9 Apr 2019

What others say about Method of Levels: How to Do Psychotherapy Without Getting in the Way →

Liked Legal Systems Very Different From Ours 13 Nov 2017

David Friedman, successfully combines the author's three special interests into a whirlwind tour of exotic law.

slatestarcodex.com 13 Nov 2017

What others say about Legal Systems Very Different From Ours →

Liked Mastering the Core Teachings of the Buddha 18 Sept 2017

The result is a sort of perfect antidote to the vague hippie-ism you get from a lot of spirituality.

slatestarcodex.com 18 Sept 2017

What others say about Mastering the Core Teachings of the Buddha →

Mixed on Raise a Genius! 31 Jul 2017

I was hoping that this book would explain Lazslo Polgar's secrets for raising gifted children. It does so only in very broad strokes.

slatestarcodex.com 31 Jul 2017

What others say about Raise a Genius! →

Recommends MealSquares 25 Apr 2017

So, if you're looking for an easy way to quit the junk food and try a low-variety diet that's unlikely to injure your hypothalamus, I recommend MealSquares as worth a look.

slatestarcodex.com 25 Apr 2017

sponsored

What others say about MealSquares →

Loved Seeing Like a State 16 Mar 2017

Seeing Like A State is the book G.K. Chesterton would have written if he had gone into economic history instead of literature. Since he didn't, James Scott had to write it a century later. The wait was worth it.

slatestarcodex.com 16 Mar 2017

What others say about Seeing Like a State →

Liked The Age of Em: Work, Love and Life when Robots Rule the Earth 28 May 2016

Note: I really liked this book and if I criticize it that's not meant as an attack but just as what I do with interesting ideas.

slatestarcodex.com 28 May 2016

What others say about The Age of Em: Work, Love and Life when Robots Rule the Earth →

Recommends Albion's Seed: Four British Folkways in America 27 Apr 2016

But I highly recommend Albion's Seed as an entertaining and enlightening work of historical scholarship which will be absolutely delightful if you don't fret too much over all of the existential questions it raises.

slatestarcodex.com 27 Apr 2016

What others say about Albion's Seed: Four British Folkways in America →
All 17 recommendations →

Where they publish

Blog Astral Codex Ten Their blog: long essays on medicine, psychiatry, forecasting and rationalism.
The successor to Slate Star Codex — long, argued posts on psychiatry, prediction, ethics and AI risk, several a week including reader comment round-ups. Has a feed.

Open astralcodexten.com
Recent
- AI Safety Is Not A Corporate Plot 15 Oct 2026
- Open Thread 455 13 Oct 2026
- Your Book Review: We Present Television 10 Oct 2026 Finalist #11 in the Book Review Contest
Show 14 more

Top Forecasters On The Plague 7 Oct 2026
Hidden Open Thread 454.5 7 Oct 2026
An Open Letter To Steven Pinker On AI 6 Oct 2026
Open Thread 454 5 Oct 2026
Our AI Midwife 2 Oct 2026 A guest post by Drew Housman
Open Thread 453 28 Sept 2026
Your Book Review: Lee Kuan Yew's Memoirs 25 Sept 2026 Finalist #10 in the Book Review Contest
The Specter Of Neuralese 24 Sept 2026
Does Georgism Work? Five Years Later 24 Sept 2026 A guest post by Lars Doucet
Hidden Open Thread 452.5 23 Sept 2026
Mysteries Of AI Generalization 23 Sept 2026
Open Thread 452 21 Sept 2026
Your Book Review: This Is Going To Hurt 18 Sept 2026 Finalist #9 in the Book Review Contest
AI Superforecasting Should Transform The FDA 16 Sept 2026 A guest post by Josh Morrison
Link verified 1 Sept 2026. Recent items update automatically from the channel.

About the English under a post

Some people here publish in a language other than English. Where they do, this site shows a machine translation beneath the post, in this typeface — the site's own, not theirs.

The post itself is never changed, moved or hidden: what is set in the serif above is exactly what the person published, and it is what to quote them on. A translation can be wrong in ways that matter, especially about tone.

Only the post's own words are translated. A quoted post, a linked article and a belief on korrents.com are left in their original language.