OpenAI Deep Research is genuinely useful – Gary Marcus

Why do people in the responses to this post and elsewhere keep talking about going beyond statistical methods? How do they know there’s something else that isn’t at some level just statistical? That seems like a big assumption to me.