Generative AI will not revolutionize game development just yet


Video creation The game requires hard, repetitive work. How could it not be? Developers are in the world-building business, so it’s easy to see why gaming industry would be excited about generative AI. With computers doing boring work, a small team can make a map the size of San Andreas. Crunch is becoming a thing of the past; games are released in finished state. A new era is calling.

There are at least two interrelated problems with this narrative. First, there’s the logic to the hype itself, reminiscent of the frenzied gold rush over the crypto/Web3/metaverse, which consciously or unconsciously sees the automation of artists’ work as a form of progress.

Second, there is a gap between these statements and reality. In November, when DALL-E was everywhereventure capital firm Andreessen Horowitz aa published long analysis on their website touts a “generative AI revolution in games” that will do everything from shortening development time to changing title types. The following month, Andreessen partner Jonathan Lai published one Twitter thread explanation aboutCyberpunk Where much of the world/text is created has allowed developers to move from asset production to higher-level tasks like story and innovation,” and theorized that AI could enable “better + faster + cost-effective” game creation. Eventually, Lai’s notes became so full of exasperating responses that she a the second thread saying “there are many problems to be solved”.

“I’ve seen some, frankly, ridiculous claims about things that are very close,” says Patrick Mills, head of franchise content strategy at developer CD Projekt Red. Cyberpunk 2077. “I’ve seen people suggest that AI can improve Night City, for example. I think we are far from that.”

Even advocates of generative AI in video games think that the industry’s hype about machine learning is getting out of hand. Julian Togelius, coordinator, says it’s “ridiculous” NYU Game Innovation Lab, author of dozens of articles on the subject. “Sometimes it seems like the worst kind of cryptocurrencies left the crypto ship when it was sinking, and then they came here and said, ‘General AI: start the hype machine.'”

Togelius explains that generative AI cannot or should not be used in game development. People are not realistic about what it can do. Sure, the AI ​​can design some generic weapons or write some dialogue, but compared to creating text or visuals, level design is atrocious. You can donate face generators with silent ears or some blank lines of text. But a broken game level, no matter how magical it looks, is useless. “It’s stupid,” he says, “you have to throw it away or fix it by hand.”

Basically—and Togelius has had this conversation with many developers—no one wants level generators that work less than 100 percent of the time. They destroy all titles and make games unplayable. “So it’s very difficult to take a generative AI that’s very difficult to control and just put it in there,” he says.



Leave a Comment