I haven't seen any negative sentiment toward Terence Tao before. Coming from outside the academic math sphere, genuinely curious if there's a real issue or if this comment is just spam itself.
This comment is spam. When Tao says something we should take it seriously.
A big trend in AI spam is to take achievements in one field that could be called "AI" and use them as evidence of advancement in other fields that happen to be called "AI".
https://en.wikipedia.org/wiki/Package-deal_fallacy
Tao has been doing a lot of demonstrations of using LLMs for search and translation by experts who already know enough about a field to judge whether generated text is valid or meaningful. Those are valid demonstrations, but they don't justify the LLM-as-intelligent-agent narrative being pushed by most of the reporting on the topic, so the whole situation reeks of payola.