Comment by sifar

The argument is a bit more nuanced than this. Historically, the failed paths on the way to proving a theorem led to newer avenues. When this is cognitively offloaded to an llm we are missing out on those developments which the process generated.