AI is still being developed in terms of accuracy. Even Google Assistant isn't that top notch yet.
For AI to be as accurate as possible, it needs high data accuracy which is why I believe many of the big tech companies are seeming trying to "invade" our privacy. Only when they do that, they'll be able to understand or rather, digitalize human behavior to create AI products that are accurate.

As per your post, AI generated comments that are happening now are not as accurate as it will be in the future.