News for hackers. Stuff that matters.
Re:Building an AI game studio: what we've learned so far by stubish on Tuesday May 21, 2024 @09:01PM (#40436213)
> haven't seen it do something that isn't already possible using a "not-intelligent" solution

This is dismissing time and resources. I think things like this game maker project are extremely impressive, because it enables an individual to do in hours what it would take a team to do in months. Just '5 different cats' is impressive, because most game studios would just do one and repeat it, because their time is limited.

I don't know if this experiment will end up producing 'real' games, probably not by most people's definitions. But I can see it being used as an amazing toy by individuals, such as kids. Generating a custom game in hours, much like they can already do in months or years using other products like RPGMaker.

Re:Building an AI game studio: what we've learned so far by oriel on Friday May 24, 2024 @03:59PM (#40469959)
I've been using godot with chatgpt as a side project to see how much I can coax a game out of it, with minimal effort. Its a manual form of systems like this but definitely shows how generative tools enable independent devs or small teams to tighten their dev loops and rapid prototype ideas.

When you can rapid prototype ideas in a game search space, its more likely to get a high quality output, or at least games that hone in on fun more effectively than a AAA made with a waterfall model and a huge team with poor crosstalk abilitiy.