my feeling is, for large games; no.

at least not generative ai.

generative ai is handicapped by goldfish tier ability to remember and keep track of things. this means for large projects it will constantly fuck things up. combine this with people not, perhaps, reading line-by-line with what ai is making, and you are more or less creating a situation where bugs, poor performance, glitches (visual or otherwise), and just shoddy craftsmanship can thrive.

for me, personally, generative ai is more or less a net plus when its dealing with things at the individual level of short scripts, functions, snippets -- probably in the level of hundreds, not thousands of lines of codes, where those things are highly-self contained, and probably serving as template or boiler-plate code.

this isn't to say that it is a net negative for ALL games, maybe for things like text adventure of visual novel games, they can be used to approach or leverage problems of exponential story branching being untenable.

but from a craftsmanship level of large games its going to be mass manufactured vs handmade, it will always be a downgrade.

and in terms of mass-manufacturing, the best possible scenario for ai assisted game development is if they go the ikea route, creating simple games, and assembling simple pieces of ai code blocks themselves.

but even then, intricate, highly optimized games with a lot of attention to detail are antithetical to current generative ai, at least as far i as feel.