Exactly this. At a recent hackathon, I spent the bulk of my time actually engineering, building custom model architecture in Lua, wrangling a dataset, and waiting on training loops. The winning project? A slick UI wrapped around a basic LLM API call.

It’s incredibly frustrating. The incentive structure at these events has completely shifted away from actual technical complexity toward whoever can build the prettiest ChatGPT wrapper in 6 hours..

They are now vibeathons. Same thing but now it's how much do you write by hand, it's how much can you delegate to the LLM while you're writing by hand.

That act of juggling has become an important part.

People will just look for harder problems to solve. It will remain but differently

Before AI people already used preexisting projects for hackathons.

Last hackathon I participated in was merely "best presentation". Code, app did not matter, it was idea and presentation what mattered.

Thankfully, properly working app won that one, nonetheless they suck now.

Betteridge's Law. Unless it's more of an essay competition for text to feed the AI beast.