These people treat math research as if it is a math homework assignment. There needs to be an honest discussion about what the LLM is doing here. When you bang your head against a math problem you blindly try a bunch of dumb ideas that don't actually work. It wastes a lot of paper. The LLM automates a lot of that.
It is actually pretty cool that modern AI can help speed this up and waste less paper. It is very similar to how classical AI (Symbolica) sped up math research and wasted less paper. But we need to have an honest discussion about how we are using the computer as a tool. Instead malicious idiots like Vlad Tenev are making confident predictions about mathematical superintelligence. So depressing.