[otabdeveloper4]

> is going to drive some incredible discoveries in all of STEM

No it won't. Natural language manipulation is never a bottleneck in STEM.

[jayGlow]

I'm fairly certain LLMs have been used to solve some novel problems already.

https://www.forbes.com/sites/anishasircar/2026/04/17/ai-solv...

[esseph]

They seem to be having at least some impact

https://www.newscientist.com/article/2527564-mathematicians-...

[jesse_dot_id]

I do not understand how you can believe that and I'm not really keen on trying to.

[HappMacDonald]

Aside from everything everyone else has mentioned, AI != LLMs

[otabdeveloper4]

There is no "AI" that isn't an LLM in 2026.