Steven Byrnes
Article
Steven Byrnes is a recurring person in the Astral Codex Ten archive, appearing 5 times across 5 issues between September 06, 2022 and July 09, 2025. The archive places it in contexts such as “Steven Byrnes on Less Wrong: I’m Mildly Skeptical That Blindness Prevents Schizophrenia”; “Steven Byrnes’ objections to the symmetry theory of valence”; “Steven Byrnes on how to reason about AI danger”. It most often appears alongside Less Wrong, Bryan Caplan, California.
Metadata
- Category: People
- Mention count: 5
- Issue count: 5
- First seen: September 06, 2022
- Last seen: July 09, 2025
Appears In
- Links For September 2022
- Highlights From The Comments On Unpredictable Reward
- Links For May 2023
- Pause For Thought: The AI Pause Debate
- Practically-A-Book Review: Byrnes on Trance
Related Pages
-
- Less Wrong (3 shared issues)
-
- Bryan Caplan (2 shared issues)
-
- California (2 shared issues)
-
- Eliezer Yudkowsky (2 shared issues)
-
- OpenAI (2 shared issues)
-
- Scott (2 shared issues)
-
- Twitter (2 shared issues)
-
- 2006 IAU vote (1 shared issues)
-
- 5HT2A serotonin (1 shared issues)
-
- 80,000 Hours (1 shared issues)
-
- 11 (1 shared issues)
-
- @itsahousingtrap (1 shared issues)
External Links
Source Context
Recovered passages from the original issue text. When the raw archive preserved outbound links inside the source passage, they are listed directly under the quote.
22: Steven Byrnes on Less Wrong: I’m Mildly Skeptical That Blindness Prevents Schizophrenia. There’s an old piece of trivia that no congenitally blind person has ever been schizophrenic (I talk about it here). Steven is able to track down a few cases of this happening, and speculates that given how rare both conditions are, maybe these few cases are all we would expect to find. Since I previously wrote about this, I’ve provisionally added it to my Mistakes Page.
Related: Steven Byrnes’ objections to the symmetry theory of valence.
Jacob Buckman on how we “aren’t close” to creating self-improving AI (yes, but we “weren’t close” to creating GPT-4 five years ago!), and Steven Byrnes on how to reason about AI danger if this is true.
Inline links: “aren’t close”, how to reason about AI danger if this is true
Nora thought that success at making language models behave (eg refuse to say racist things even when asked) suggests alignment is going pretty well so far. Many other people (eg Rafael Harth, Steven Byrnes) suggested this would produce deceptive alignment, ie AI that says nice things to humans who have power over it, but secretly has different goals, and so success in this area says nothing about true alignment success and is even kind of worrying. The question remained unresolved.
Inline links: Rafael Harth, Steven Byrnes
Steven Byrnes is a physicist/AI researcher/amateur neuroscientist; needless to say, he blogs on Less Wrong. I finally got around to reading his 2024 series giving a predictive processing perspective on intuitive self-models. If that sounds boring, it shouldn’t: Byrnes charges head-on into some of the toughest subjects in psychology, including trance, amnesia, and multiple personalities. I found his perspective enlightening (no pun intended; meditation is another one of his topics) and thought I would share.