Will LLMs Autonomously Solve Math Problems in 1-3 Years? Developer Says Directed Prompts Are Just Coping

shakoistsLog · x · 2026-07-21

Addressing the current limitation where language models (LLMs) still require directed prompts to solve complex mathematical problems like the Jacobian conjecture, the author offers a contrasting perspective. The author argues that the belief that models still need human guidance is merely "cope." They predict that within the next 1 to 3 years, LLMs will advance enough that simply instructing them to autonomously find and solve math problems will be entirely sufficient.

Original post →

More from AGI Musings

AGI Musings channel →