News for hackers. Stuff that matters.

A comment on There's never been a better time to go into math (by: 3blue1brown) [video]. Parent comment

Re:There's never been a better time to go into math (by: 3bl... by eddieplan9 on Sunday October 18, 2026 @09:07PM (#50147180)
I suggest listening to "Math Vs Machine" [1], which has a much more nuanced take. It clearly lays out two types of use of AI in math, one that helps humans have better understanding, and one just bulldozes problems without helping us get any insights. It also touches on the topic of "does it matter if the machine does things in ways humans cannot understand".

[1] https://radiolab.org/podcast/math-vs-machine

[ Parent | on HN ]
Re:There's never been a better time to go into math (by: 3bl... by ericd on Sunday October 18, 2026 @09:39PM (#50147379)
After it bulldozes the problem, can you not then use the LLM to help you understand the solution, bit by bit?
[ Parent | on HN ]
Re:There's never been a better time to go into math (by: 3bl... by jimbokun on Sunday October 18, 2026 @10:03PM (#50147506)
Not clear.

Mathematicians are saying the current papers are pretty inscrutable and use terminology unfamiliar in the math sub fields in which the models are generating proofs.

[ Parent | on HN ]
Re:There's never been a better time to go into math (by: 3bl... by zer00eyz on Sunday October 18, 2026 @11:31PM (#50148066)
Cause it looks like it's mining one subfield for a pattern and applying it to another. (The take from the few math folks I know who have looked into items of their interest).

Computing has the same issue. If we swapped a PLC programer with your average SV software engineer for a week both of them would have a really bad time.

[ Parent | on HN ]
Re:There's never been a better time to go into math (by: 3bl... by ericd on Sunday October 18, 2026 @11:54PM (#50148204)
Good thing LLMs are really good at translation tasks?
[ Parent | on HN ]