Feb 23, 2023 - Technology

Game makers put generative AI to imaginative work

Illustration of a distorted wireframe landscape with geometrics shapes around it.

Illustration: Lindsey Bailey/Axios

Video game makers are already trying to use advances in generative artificial intelligence to build virtual words faster and fill them with life.

Why it matters: Game creation is full of problems that generative AI can potentially solve, and in the process make a very public showing of the tech’s potential.

One dream that's now in rapid, real-time testing is to harness generative AI to put more words in the mouths of the dozens or hundreds of computer-controlled characters that a player might meet in a role-playing game.

  • Typically, even the biggest budget games with hundreds of developers on hand can include just a few lines of dialogue for most of the incidental characters in its world.

What's happening: A new, fan-modified version of the role-playing game Mount & Blade II is taking a stab at having ChatGPT-powered citizens who can say more things to the player

  • “Neat,” wrote Techspot — with the caveat that AI-generated replies for this experiment took long to load and sounded somewhat robotic.
  • The upcoming Origins for the PC, which the developers say is not a game but a “technical demo,” is offering a similar test, offering AI-generated responses from the computer-controlled characters as they are interrogated by the player.

Other developers are dabbling with generative AI to create game code or the visual building blocks of new virtual worlds.

  • Independent game developer Alex Rose tells Axios that he’s been asking ChatGPT to help him write code for Unreal Engine, after he trained for much of his career on competing tools. “It has an 80% success rate for me,” he says. “20% of the time it’s kind of gibberish.”
  • During an Axios visit to Sony-owned Haven Studios in Montreal last fall, developers there said generative AI could help create rough virtual characters and objects based on simple text prompts. That could empower more members of Haven's team to visualize ideas they have for the game, and free skilled artists from mundane tasks so they can spend more time on visual refinements.

What they’re saying: EA co-CTO Marija Radulovic-Nastic recently told Axios that AI would have “a big impact on the way we make games.”

  • It can help with concept testing, help draw out and fill game levels, build avatars and add voice-work for user-made content, and more.
  • But people shouldn’t expect AI to make broadly appealing games on its own any time soon, she added.

“One of the biggest challenges in game making is getting to the fun fast,” she said, describing the painstaking process of building and testing gameplay ideas.

  • Generative AI that can help build prototypes faster could free designers to distinguish bad ideas from good ones more rapidly, and maybe make the good ones great.

The big picture: Most of the generative AI concepts for gaming amount to easing the pipeline congestion that makes game creation so time-consuming and labor-intensive.

  • That’s the thesis behind Roblox's announcement last week that the company is developing proprietary generative AI tech meant to empower its users to make virtual objects for Roblox with simple text prompts.
  • Currently the company relies on a small number of its users to wield its coding language to create games for the rest of its 59 million daily active players to experience.
  • Generative AI “tools can not only accelerate creator productivity, but can significantly drop the technical skill needed to bring ideas to life,” CTO Daniel Sturman wrote.

Yes, but: The deployment of AI to make game creation easier has its hazards.

  • Widespread use of generative AI may exploit creators if their work is used by the AI without permission.
  • This moment in the buzzword spotlight for the new tech might fade fast, as happened with the NFT craze.
  • And then there’s the anxiety about AI replacing people. Rose, the programmer, isn’t sweating that last one: “I'm not yet afraid for my job,” he said. “I'm more excited for what it can do for my job.”

The bottom line: Use of this kind of AI in games will test whether it’s possible to move the horizon of what can be created by people and machines working in concert.

Go deeper: Read more in Axios' AI Revolution series —

Editors note: This story has been updated to clarify that Origins’ developers consider it a technical demo, not a game.

Go deeper