@jaredwhite@indieweb.social @davidgerard@circumstances.run There is a corresponding pattern in AI and AI adjacent computer science research. At various times since the 1950s, all of the following were said to be on the cusp of matching or exceeding human intelligence and ushering in the age of The Terminator or The Jetsons: neural networks; expert systems; case-based reasoning; cybernetics; the subsumption architecture; reinforcement learning; neural networks again; recurrent neural networks; reinforcement learning again, this time with neural networks. You have to pose as though you believe if you want to remain a member in good standing. Phil Agre has written about what happens when you openly express that you've lost the faith.
When I was a graduate student in the early 2000s AI agents were all the rage too. Now here we are in 2025 with "agentic AI", the same stuff except with LLMs instead of logic programming in the innards. Somehow we're still pretending that the stuffing in the animatronic toy is what determines whether it's intelligent.
Incidentally, back then I saw a talk that, roughly summarized, argued "quantum computing can't work" and here we are, in 2025, with folks still arguing quantum computing is just about to pop even though the problems from the early 2000s haven't been addressed (as far as I know--not my area of expertise--but I still see papers reporting work on those same problems). So that's fun!