Models are already smarter than any expert — use that instead of arguing
15 episodes in 12 months · 25 total · first in September 2024
In volume of knowledge, speed and quality of analysis, modern models have already overtaken any Nobel laureate. Codex and Claude Code write code and think through architecture better than 99% of programmers. You can argue as much as you like about whether AGI has been reached, but honestly — the current model is smarter than any human.
So I trust AI more and more where I used to go to a doctor, a lawyer or an accountant, and I check their work through it. People badly underestimate how far AI has come and use it for small things.
Already happened. For me this is already a fact, not a forecast: in No. 166 (September 2026) — “the current model… is smarter than any human”, and in No. 148 — Codex and Claude Code write code better than 99% of programmers.
Where to start
Before going to a specialist or agreeing with one, work through the question with a strong model.
Check the work of lawyers, accountants and contractors through AI.
When I said it
Quotes
“…its level is very strong. Right? I mean, it's comparable to a PhD, to the intelligence level of a PhD person.”
“Because today's system, in terms of, for example, the volume of knowledge, obviously, modern artificial intelligences, models like ChatGPT or Anthropic, they've outstripped a human in knowledge by a long way, yes, any Nobel laureate in knowledge, in speed, in calculations, in approaches and in everything else.”
“I believe, better than ninety-nine percent of people in the world. I'm talking now about programmers. Better than ninety-nine percent of programmers in the world they test, write code, analyse, work out the architecture, systems analysis and so on.”
“It's like these statements that artificial intelligence still hasn't reached AGI, because AGI is, like, a superhuman, but let's be honest here, the current model of artificial intelligence is smarter than absolutely any human, smarter than any human.”