I just don't think you can ad-spam LLM results as effectively as traditional search results, because they're so much better at giving you the answer you want. Vs normal search you ask a question and you're forced to dig through a few links and back out when you inevitably can't find what you're looking for.
Although maybe the GPU economics won't work unless you ad-spam and the other services will offer similar amounts of spam down the road... will be interesting to watch for sure.
We could be in a situation, again for a while, where "really good advertising" becomes isomorphic to "useful information." I can see it in most of my GPT interactions -- I would welcome links to books, products, courses, etc., relevant to the "discussion" I'm having w/ the LLM. Seems feasible to do, but we shall see.
reply