Stephen Boyle

[Stephen Boyle @Stephen_Boyle]

@bbclaurak @elonmusk None of the three steps in the actual article involve any firm commitment to slow down.

[Stephen Boyle @Stephen_Boyle]

we should make norms or laws against this in the same way we do animal cruelty.

[Kevin Ngo @kevin_t_ngo]

I've trapped the fruit fly in my rabbit r1.

When I shake it, I can see its brain's escape circuit light up

[Stephen Boyle @Stephen_Boyle]

Wait until everyone just learning about AI risk learns about Roko’s basilisk

[Mick West @MickWest]

This time is different.

I think it's a lazy mistake for skeptics to equate warnings about AI's dangers with historical doomsday predictions.

I think we'll probably still have human civilization in ten years. But a bomb not going off doesn't mean it isn't dangerous. If AI does not cause significant and dangerous global chaos, it will be because people have heeded these warnings.

I've worked extensively with AI for the last two years. The increase in its capabilities has startled me. It's progressing very fast, in unexpected ways, and there's nothing holding back that progress.

That does not mean the bomb is going to go off. But it does mean that we, humans, can't ignore it.

Are some of the warnings hyperbolic? Perhaps. But they still reflect a real danger. Civilization-level danger. Good people need to be working on mitigating that danger.

So chuckling about how end-of-the-world predictions have never panned out in the past not only misses the true nature of the challenge ahead; it might also make things worse.

[Packy McCormick @packyM]

look i get that extinction is scary, but what's the alternative? solve our own math problems?

[Stephen Boyle @Stephen_Boyle]

@Peston He wrote this in 2023… https://t.co/T3eW5vCYVQ We must slow down the race to God-like AI

[Joe Weisenthal @TheStalwart]

The big overlap is “AI safety skeptics” and “people who thought crypto would be the next big thing in 2021”

[Evan Hubinger @EvanHub]

Jacob is correct here—we really do earnestly believe AI could kill all humans! I personally think it is >10% within the next decade. I believe Anthropic is trying its best, but we do not yet have a plan to solve alignment for superintelligence and are not clearly on track to.

[Jacob Coxon @hilbertspaess]

I resigned from Anthropic today. I spent the last three years doing pretraining research at both OpenAI and Anthropic. Neither company is acting responsibly. They are racing straight to self-improving superintelligence and gambling with our lives. More thoughts below.

[Izabella Kaminska @izakaminska]

This is absolutely fantastic and super creative journalism from @jimwaterson. I’m utterly gripped. TLDR: London’s tourists are being fleeced by a tightly controlled illegal ice cream van cartel/syndicate which is operating in plain sight outside the Houses of Parliament.

It’s like the plot of Wonka but with ice cream vans and cute chime-based melodies, instead of a chocolate cartel.

Makes vape shop look positively legitimate in comparison.

https://t.co/q5ebt76eUJ?

[Sebastian Payne @SebastianEPayne]

Incredible letter in @thetimes

[Neil Irwin @Neil_Irwin]

Chart of the year:

[Stephen Boyle @Stephen_Boyle]

@dkennytcd @MatthewHolmesBL For reference https://t.co/723MMzBf4i

[Stephen Boyle @Stephen_Boyle]

@dkennytcd @MatthewHolmesBL Interesting! I’m starting to find the search augmented retrieval / extended thinking models (eg GPT 5 Thinking) can get alright results (for the original question pulled back a TCD paper that directly states the number of referrals - though your colleagues may have been wrong 😉)

[Stephen Boyle @Stephen_Boyle]

@donlav @StuartWelsh4 AI generated union flag as well... unless the video was taken pre 1801 when the St Patrick's cross was added...

[Stephen Boyle @Stephen_Boyle]

@nntaleb @IndianLatino This is why market orientation (exposed to competition or not) was predictive of post-soviet success for businesses while ownership (public or private) wasn’t.

[Stephen Boyle @Stephen_Boyle]

@doctorow Some good ideas here. However to say the new state lender could always get their money back because of repossession (foreclosure) seems a stretchin when combined with the price depreciation element of the PILL? Interest isn’t just profit for banks - it also covers default risk

[Stephen Boyle @Stephen_Boyle]

Ah yes. Reading comprehension in the age of LLMs. Who would need that?

[Celia Richardson @CeliaRichards0n]

I’m Spartacus. English Lit.
Je ne regrette rien (sorry I do realise that’s French).

[Stephen Boyle @Stephen_Boyle]

@StartupArchive_ It’s so very Thiel that he’s confidently saying this about a book Gladwell didn’t write.