Check, what François Chollet had to say about the current hype and lack of data in the AI field is reminiscent of the early days of web3, with high expectations for civilization-altering impact and returns on investment in the near future
Preview: The writer discusses their thoughts on the current state of AI and how it relates to the web3 space, noting that there are similarities in the way that narratives are accepted as self-evident without data to back them up. They present a "bull case" and "bear case" for the future of AI, with the bull case being that generative AI becomes a widespread user experience paradigm for interacting with most tech products and the bear case being that large language models (LLMs) only find limited success in specific niches. They argue that while there is currently more evidence supporting the bear case, they believe that LLMs will improve in the next few years and that the actual outcome will likely fall somewhere in between the two scenarios, with AI becoming a universal interface to information in the future but not quite happening with this generation of technology. They also note that any successful scenario will likely be undercut by commoditization and point out that so far, image generation is the only proven commercial success for AI.