I do think that segment missed a lot of crucial points, and focused on very near term issues that will no doubt be overcome relatively easily.

But The hallucination aspect has to be solved and it needs to happen very soon. I think once that is tackled the train won't stop. I think it will be reduced over the coming months to the degree that it becomes a non-issue in most cases fairly soon. Google has a lot riding on that.

We also shouldn't underestimate how much more useful a model with access to the internet will be over the current chatGPT. Recent events will prove very useful.