Henri Rapson

NHS policy lead by day, vibe coder by night. Interested in AI policy, safety, and broader effects on society.

[Henri Rapson @Hrap]

@AlexHormozi What if the 'hard' path only seems hard because we're thinking only of the costs, not the outcomes/benefits, of taking action? Wouldn't pushing through simply reinforce that incorrect habit of thought?

[Henri Rapson @Hrap]

@MsMelChen Three years and still no visits to @Tesco, @sainsburys or @Waitrose? Sad times indeed.

[Henri Rapson @Hrap]

@JohnRentoul Blair? Don't you mean @George_Osborne?

[Henri Rapson @Hrap]

@jun_song I wonder if this generalises, indicating that their abilities are far more profound than they estimate? Claude and Codex might be suffering from self-limiting beliefs.

[Henri Rapson @Hrap]

@thetreygoff If you are using Opus 4.8 Max against 5.5 with Heavy thinking, that's not an apples to apples comparison.

[Henri Rapson @Hrap]

@thetreygoff Thanks, so GPT 5.5 Pro then on the GPT side?

[Henri Rapson @Hrap]

@foundmyfitness 15 people in the sample 👀

[Henri Rapson @Hrap]

@ClaudeDevs Thank you!

[Henri Rapson @Hrap]

@RichardDawkins Thomas Nagel has entered the chat....

[Henri Rapson @Hrap]

@deanwball The original o3 was a wonderful model, the first time AI felt close to agentic. It is sorely missed.

[Henri Rapson @Hrap]

@minhsmind 3.1 Pro beats almost every other model - including Mythos - across several benchmarks. Gemini is also the only natively multimodal LLM. So Google is likely building for a huge future inflection point. And, remember, they own part of Anthropic, and seem generally supportive of it.

[Henri Rapson @Hrap]

@loganclarkhall The letters which survive are not representative of the vast majority of their society. They are from those who were, by definition, capable of writing. And those historians selected to preserve. It's an unfair comparison.

[Henri Rapson @Hrap]

@Oxandrolonely @lietotajvardsj @BourgeoisieH8er @NBTiller It also doesn't really matter if he did, because they're not mutually exclusive. If I'm driving from A to B in a Porsche, does that mean that I shouldn't recommend that you buy a Ford for the same journey?

[Henri Rapson @Hrap]

@NBTiller @Oxandrolonely Also, @hubermanlab is incredibly rigorous and precise with any claims he makes. No credible scientist - let alone a Stanford professor - would ever claim that any health outcome is strictly unifactorial.

[Henri Rapson @Hrap]

@NBTiller @Oxandrolonely He disclosed this several years ago, close to when he started the podcast.

[Henri Rapson @Hrap]

@claudeai @Grok can't Claude Code do this already?

[Henri Rapson @Hrap]

@jtregunna @joseph_h_garvin This is correct. But with the amount of errors and bugs I find in every single output, the loop element here is dangerous. These systems still need thorough reviews.

[Henri Rapson @Hrap]

@robinhanson As @nntaleb pointed out, in an almost equivalent hypothetical in The Black Swan, you need to compare pre and post exposure to know if this is true. The survivors may be weakened compared to their pre-exposure selves.

[Henri Rapson @Hrap]

@robinhanson @grok were the radiation exposed subjects weaker in any dimension compared to before their exposure?