AI Could Improve Games by Clearing Old Code and Expanding Creative Options

For Bing Gordon, AI’s near-term value in game development may be less about producing a finished creative vision than taking on work that slows teams down. He points to “tech debt”: old, scruffy code that engineers may not want to revisit. If it is not reformatted, Gordon says, development slows down. AI could help clear that backlog, with the result showing up in higher-quality games.

He sees a second benefit in the number of ideas teams can explore. The best creative people, he says, may come up with 100 ideas in the time it takes others to come up with three. AI could help teams generate and test ideas faster. Gordon expects higher quality “around the edges” from those additional attempts.

The two benefits address different constraints. Working through neglected code could reduce slowdowns; generating and testing ideas could give creative teams more chances to find something that works. These are practical uses, distinct from Gordon’s longer-term ambition for generative AI: characters as interesting as the people he admires. He imagines having dinner again with his late mother, or meeting Abraham Lincoln the night before the Gettysburg Address. He describes wanting to “travel through time and space and drama.”

Asked whether AI’s gains would favor nimble startups or established studios, Gordon did not name a particular bet. He said developers may use faster work to pursue more original ideas, including ones that initially seem strange. As an investor, he has learned that laughter at a company idea can sometimes mean it is unusual enough to work. He cautions against shutting down exploration too soon.

His example is a “rocket buggy” developed while he was working on a game with Mark Skaggs. Gordon thought the vehicle could become a game in its own right and urged Skaggs to work on it for two weeks. The next day, Skaggs said he could not: he had to finish all the art within a year. Gordon calls that kind of time pressure an enemy of quality.

He also expects game makers to become “AI prompters.” Part of the opportunity, as he describes it, is reducing communication breakdown when one person tries to convey a creative vision to another. A designer passes an idea to a programmer, who passes it to an artist, who passes it to a musician; Gordon says things can break down along the way. Drawing, music and body language help, but some teams collaborate better than others. He sees potential in reducing the “phone game” of passing a vision from person to person.

Gordon’s account of game quality also puts a limit on what faster tools can accomplish. Quality comes in layers: a team builds something, looks at it, and builds something on top of it. Yet he says that for most creative people, delays make things worse. In a business shaped by Moore’s Law, missing a release by a year can leave the art and technology feeling old. Only a few creators, in his view, can use extra time to make something better.

Gordon points to Rockstar and Sam Houser as a team that keeps adding layers instead of settling when others might. He describes that persistence as a “Steve Jobs lesson,” and says Rockstar has shown repeatedly that it can keep building. Some things, he says, are worth waiting for; he hopes people enjoy the next Grand Theft Auto as much as he expects to.

Gordon describes video games as the largest form of entertainment by dollars and hours, larger than he and his colleagues expected when they founded Electronic Arts in 1982. His book, Everybody Wins: What Games Teach Us About Making Products People Love, written with Chloe Gordon, draws on lessons from making games that sell and please people and applies them to other industries.

His AI examples sit at different distances from current development work. He describes clearing neglected code and generating and testing more ideas as ways AI can help teams move faster. His hope that generative AI will create compelling characters is more personal and speculative: he wants to encounter people from his own life and from history through characters that hold his interest.