Income and life expectancy are strongly correlated. However, studying
Swedish lottery winners (a natural experiment on increased wealth) finds no
causal effect.
"At this point I knew that we had a counterexample — a group scheme of
order 4 which was not killed by 4. I suggested to Akhil that he make a PR
to mathlib with the counterexample — which he did. I would have also
suggested to him that he draft a press release saying that a machine had
solved a 60-year-old question of Grothendieck in algebraic geometry, but
somehow by this point I was almost becoming immune to all of this. It
wasn’t clear to me that the media would even be able to distinguish
between “machine resolves question due to Erdős” and “machine resolves
question due to Grothendieck”."
Lots of advances in mathematics made by AI this month (probably more than
humans?). Because it doesn't require real world experiment, math is gonna
see some breakneck progress. What a time to be alive indeed.
Plan A
Larsen, Dean, Halstead, Lifland, Greenblatt, and Kokotajlo, July 2026
A mix between a forecast and a policy prescription on how the U.S.
will/should address powerful AI systems. Plenty has been discussed on this
so I won't add too much. Like others, I dislike the muddied distinction
between normative policy recommendations and predicted unfortunate actions.
I also think they overplay job loss (cf.
Alex Imas)
and single-agent theory. Finally, I'm skeptical about slowing progress by
limiting algorithmic improvements; theory/algorithm improvements are exactly
what will be accelerated most by powerful AI (cf. "Human mathematicians are
being out counter-exampled") and it's hard to know a-priori the impact of
fundamental research. Alignment research can advance capabilities (e.g.,
RLHF) and theoretical research (e.g., learning mechanics) could as easily
advance capabilities as improve alignment.
Rising status: floating datacenters, datacenter hardware for distinguishing
inference from training workloads.
Read
They're Made Out of Meat
if you haven't already. If you've read that post you could already guess
this one's whole trajectory. Regardless, very funny.
Agree with Noah's framing of decreasing returns on intelligence. In fact, I
wrote a very similar blog post of my own earlier this month on
The Limits of Intelligence.