> I think Google, with their massive YouTube data set, is ultimately going to win this game.
I don't know, applying the same thinking to LLMs, Google should have been first and best with just text based LLMs too, considering the datasets they sit on (and researchers, among others the people who came up with attention). But OpenAI somehow beat them on that regardless.
The problem for Google existed with the infobox at the top of search results. If users get the answer to their query without having to visit the web page where the answer came from, and where Google shows the ads, means that users don't see ads, and that website operators don't get ad revenue. ChatGPT was Google's Kodak digital camera moment. They had internal transformers-based chatbots (that really wanted to send you pizza, for some reason), but deploying that would have cannibalized their existing business model, so in the meanwhile, their lunch got eaten by an outside competitor.