> unseat ChatGPT with a market-leading general purpose assistant.
It's not impossible. The prediction from many(not that I believe it) is that over long run modelling tricks would become common knowledge and only thing that matters is compute and data, both of which Meta has.
Also there could be a trend of LLMs for ads or feed recommendation in the future as they has large completely unstructured dataset per user across multiple sites.
Compute, data, and most importantly distribution/users.
IMO standalone AI companies like OpenAI might be successful by providing infrastructure to other companies, but I can’t imagine ChatGPT remaining #1 many years from now.
The web is still trending towards being a walled garden. Maybe not right now, but long term I think people will use whatever AI is most convenient which probably will be AI built into a giant company with established user base (FB, GOOG, MSFT, and Apple if they ever get around to launching - would love Siri 2.0 if it meant not needing to open the ChatGPT iOS app)
It's not impossible. The prediction from many(not that I believe it) is that over long run modelling tricks would become common knowledge and only thing that matters is compute and data, both of which Meta has.
Also there could be a trend of LLMs for ads or feed recommendation in the future as they has large completely unstructured dataset per user across multiple sites.