🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

We desperately need a better term than "AI for science".

Maybe the meta-lesson is to be less credulous about single-author papers by grad students that aren't self-verifying (like proofs, etc.)

Everybody wants to be like VCs and Silicon Valley except in perhaps the most important ways: - Speed - Relatively large chunks of unconstrained funding - Delegation to operators - Healthy disrespect for seniority

Automating experiments with AI is going to be really hard because automation is way less about high-dimensional Bayesian optimization than it is about figuring out why there are bubbles in your syringe pump line

This generalized version of Mike Solana's air travel prediction is what I fear is the default path for most things unless we continuously work to maintain, improve, and make the world more awesome. We should fear civilization not ending with a bang, but with a whimper.

The most painful part of writing is realizing you don’t quite know what you actually think

The corollary is that efficiency is overrated. (As in so many cases.)

The world ends all the time. That is, *your* world -- the world you're comfortable in, the world you understand, the world you can plan for -- will definitely end quite soon, no matter who you are. Definitely since the Industrial Revolution, maybe in a different way before that.

This is not what I meant by "unbundle the university"

The move is to create a new paradigm around tooling-free manufacturing. To a large extent, the pieces are there — we need to extend them and figure out the whole system.