Re:Part two of Grant Sanderson's video with Terry Tao on the...
by 7thaccount on Thursday February 27, 2025 @05:00PM
(#43199007)
I'm not so sure. It seems these kinds of people are exceedingly rare that grasp a fundamental truth we've all missed. ChatGPT is really cool, but building a really really big neural network and training it off all the stuff on the internet certainly seems less impressive than discovering calculus. I mean ..LLMs are impressive and magical of course.
You need a certain amount of motivation, grid, etc. If these skills would be broadly available and the situation also, we would have a lot more Einsteins today.
There is plenty of things which do not require global projects building multibillion devices.
Mathematic thinking for ML for example.
DeepSeek paper for example.