It is raw research data though. Human math is way more refined. LLMs are superhuman at deduction, yet cannot create their own pre-training (where most learning happens) data.
{"deleted":true,"id":50254599,"parent":50252975,"time":1793073939,"type":"comment"}