If we limit ourselves to a human directing some sort of AI-driven code generator, if it can produce a Wolf3D or DooM-type game at present, we'll see an EQ-style game within ten years, possibly before this decade is out,. That's especially likely if you allow for an experienced human, not just a total programming newbie. MMOG's are complex, but broken down into their component parts--host, client, network code, database, graphic engine, textures, etc--it's manageable. You're just replacing the junior coders with automation. We'll get there.
If you want to simply provide "OmniAI" a prompt of "produce me a circa 1999-style multiplayer virtual world with a medieval fantasy setting" then it spits out a fully completed game a couple hours later, I think we're farther away for the reasons discussed. AI's an oxymoron term, it has no genuine intelligence, and as such it can handle math or produce functional code--known things with a logical basis--has no concept of things like enjoyment or fun. Your AI-generated game would be, to use the current buzzword, so much "AI slop," perhaps functional but likely soulless and unenjoyable.
Same as automation in a lot of other industrial fields, really: The more reptitive and lower-skill positions end up replaced by machines, but you still need skilled operators keeping the overall system functional.
Thinking about it, if AI gets to the point where it can also replicate the "human factor," I don't think a 1990's style video game will matter because it'll by extention be able to do so much more.
|