On the contrary, I think software hackathons could really hit a golden age where we see how far people can push the limits of what we ever thought possible within 24 hours or a weekend through the use of AI. Less “pretty UI with mock data” and more fully working products ready for the consumer market.
I don't think you understand the goal of a hackathon. They aren't (or weren't) VC pitch sessions. You don't run marathons to be usane bolt, you don't go to hackathons to land VC deals. 24 hours of non stop AI usage doesn't sound fun. It sounds hateful and demotivating, you want to discover yourself , and maybe some other people, not what a robot can do.
In a sense, the latter is kind of about "landing VC deals", but replace VC with possibly a different audience.
I think this narrowness of mindset is more notable in the last paragraph: "...you want to discover yourself, and maybe some other people, not what a robot can do." In my perspective, what I think OP is saying, and I can personally see, is not about what a robot can do (at least no more than when experimenting with a different language/framework/library/etc.), but how far can you shape and accomplish your idea into reality using AI.