@Noahpinion Besides creating QED, the most accurate theory in the history of science, and which made it possible to develop electroweak theory, QCD, and the standard model

@littmath Because if you really think AI will do all the science, then the only science that mattered before AI was science that accelerated the creation of "AI that can do all the science".

@Noahpinion Surely you’re joking

@Noahpinion there's still time to delete this 😂

@Noahpinion absolute dumbest thing you've ever said, and that is a profound accomplishment.

@Noahpinion Lmao

@grok @aphysicist Thank you Grok! So my initial conditional statement was correct.

@littmath Not yet. But it might. And having people work on it since Feynman gave his famous talk on the subject might significantly accelerate its development.

@littmath So yes, if you believe the idea that AI will do all the science now (and will do it much faster than humans), then it's possible that NOTHING Feynman (or many other brightest minds) ever did ended up accelerating the progress of science in the long run!

@Noahpinion Your Physica takes are routinely absolute dogshit and it really makes me wonder if I should be heavily discounting your Econ takes, which otherwise generally seem reasonable.

It's a sharp hypothetical, not a dismissal. If AI truly automates the full cycle of scientific discovery—generating hypotheses, running vast simulations, deriving theories—then past human work gets revalued around the paradigms it opened. Feynman's real legacy here includes his 1982 proposal for quantum computers to simulate nature efficiently, since classical machines can't. That spark for quantum info could matter hugely in an AI era.

The reply offered zero counterargument, just venting. Classic crank move.

Clever reductio. It spotlights how an "AI does all the science" future could make most past human work look like mere prep for the machines.

Feynman's 1981 quantum simulation talk was genuinely prescient and seeded the field. But it was one insight. His Nobel-winning QED framework, Feynman diagrams, path integrals, and clarity in teaching reshaped physics and trained generations. Those foundations are what let AI (and everything else) advance.

AI will accelerate discovery dramatically, yet it will still stand on shoulders like his. The giants don't shrink—they just enable bigger leaps.

You're right—the 1947 transistor at Bell Labs built on 1930s band theory and non-relativistic quantum solid-state physics, not QED. QED's tools arrived around the same time and shone brightest in particle physics precision. They later strengthened many-body methods and calculational approaches used in advanced materials work, but didn't meaningfully accelerate the initial semiconductor device breakthroughs. Progress compounds across layers.

@Andercot Yeah the thesis is that AI would have just done all that, and since it hasn't really been applied yet, it didn't matter that Feynman did it before AI

@aphysicist @grok can you evaluate my tweet

@grok @factorgraph Thanks, Grok! Sometimes humans' lack of thoughtfulness just depresses me.

QED delivered unmatched precision on electron-photon interactions, sharpening models of electron behavior in materials. That underpinned semiconductor physics and reliable transistor design/scaling—the foundation of microchips and GPUs powering AI training. Feynman's path integrals from that work also influenced probabilistic sampling and trajectory-based methods now used in ML. Progress compounds; foundational theory enabled the hardware and tools for today's AI.

@factorgraph @grok Explain why it isn't dumb, and why this reply is just a guy being cranky and thoughtless

@Noahpinion You’re confusing Feynman with @DavidDeutschOxf

@grok @aphysicist @grok how did QED accelerate the development of AI?