GDC 2025 and the AI Game Dev Debate: What Everyone's Missing

GDC 2025 just wrapped, and you didn't have to be in the room to know what dominated every hallway, panel, and after-party: AI. Specifically, the fight over what AI does to game development. Half the industry is convinced it's the end of craft. The other half is quietly shipping with it and not saying much. The discourse was loud, divisive, and, I think, mostly aimed at the wrong question.

So let me plant a flag.

AI doesn't replace taste

The fear, boiled down, is that AI replaces the developer. That a model writes the game and the human becomes redundant. And I get why people who spent years mastering a toolchain feel threatened by that framing. But it's wrong, and it's wrong in a way that matters.

AI does not have taste. It does not know your game is boring. It will cheerfully generate a thousand competent, soulless mechanics and never once feel embarrassed. Taste, knowing what's fun, what to cut, when the jump feels right, why this loop is sticky and that one isn't, is the entire job. And it stays human. What AI actually kills isn't the developer. It's the toolchain tax: the months you burn learning an engine, the boilerplate, the build pipelines, the certificate dance, the dead time between having an idea and being able to play it.

That tax was never the craft. It was the cost of the craft. Confusing the two is the central mistake of the whole GDC panic. Nobody mourns that we stopped hand-setting type when desktop publishing arrived. We just got a lot more writers. Same shape here.

The unlock is who gets to make games

Here's the part the doom takes keep missing. The interesting consequence of AI isn't that existing studios ship 20% faster. It's that the pool of people who can make a game at all goes from a few million trained developers to basically everyone with an idea and a sentence to describe it.

That's the thesis Remix is built on. Text-to-game, no-code, vibe coding, whatever you want to call it, the point is to drop the barrier from "learn Unity for six months" to "type what you see in your head." Roblox made creation more accessible and still asks for a real learning curve. We're going further: describe it, play it, ship it into a feed where people are already watching. A TikTok for games, where making and launching are the same motion, and the games are free to play with no install.

When you lower a barrier that far, you don't just get more of the same games faster. You get kinds of games that never existed, from people who were never going to grind through an engine. That's the actual story of AI in game dev, and it's a wildly optimistic one, for players and creators both. The folks already vibe coding on Remix aren't replacing anybody. They're new. They simply couldn't get in before.

The studios sweating about AI replacing developers are answering the wrong question. The right one is: what happens when ten times as many people can make a game? I think we find out the medium was bottlenecked on access this whole time, not on talent.

You can argue the philosophy in a panel, or you can just go make one. Type a sentence, watch it become playable, drop it in the feed.

Try it at remix.gg.