Strauss Zelnick says artificial intelligence will never be able to match Rockstar Games’ franchise.

Although AI can streamline parts of game development, Zelnick argues it would never create a truly great Grand Theft Auto game because current models lack something essential. Speaking about AI in game development at CNBC’s Technology Executive Council Summit in New York on October 28, he made it clear that today’s models could not deliver a high-quality Grand Theft Auto. Using AI to generate intellectual property also raises questions about protecting one’s own assets and respecting others’ rights.

According to Zelnick, even without restrictions on AI, you couldn’t push a button tomorrow and produce a plan on par with a Grand Theft Auto marketing strategy — let alone the game itself. The answer, he says, is no. It isn’t possible yet, and nothing truly good would come of it. Ultimately, you’d get something derivative. That lack of originality is what prevents AI from making an excellent Grand Theft Auto. Predictive models trained on historical data can work well in certain areas, but they aren’t capable of crafting the multilayered universes in Take-Two’s games.

He added that Rockstar’s team creativity is extraordinary, and the studio’s goal — achieved many times before — is to build experiences that approach perfection. By definition, AI models aren’t creative because they’re data-driven. While Zelnick’s case is compelling, the industry remains divided on AI. Electronic Arts, for instance, actively encourages employees to use AI across tasks, from coding to QA.

Others — such as Super Smash Bros. creator Masahiro Sakurai and Metal Gear and Death Stranding creator Hideo Kojima — view AI as a tool that could sustain large-scale development and enhance creativity in the future.

Source: WCCFTech, CNBC