With Rockstar Games still listing GTA 6 for 19 November, the conversation around its development has returned to AI. In a new interview with The Wall Street Journal, Take-Two Interactive CEO Strauss Zelnick described Grand Theft Auto as handcrafted while discussing the company's interest in using technology to make better games more efficiently.

It's a useful conversation to have around this particular game. GTA 6 puts enormous attention on the people building its world, while its publisher remains interested in AI's practical uses. The interesting part is how those tools fit around a team with a strong idea of what it wants to make.

The Handcrafted Promise Hasn't Changed

In its 5 November report on the interview, Insider Gaming quotes Zelnick saying, “Grand Theft Auto is handcrafted.” He also points to the industry's long history with AI and machine learning, and expresses confidence that technology can help the company improve its games and efficiency.

There is important earlier context. On 3 February, Zelnick told GamesIndustry.biz that generative AI had “zero part in what Rockstar Games is building” for GTA 6. He described the studio's deliberate construction of its worlds, while separately discussing AI experiments and tools across Take-Two.

That February statement is still worth reading alongside the new remarks. This week's interview doesn't identify generated artwork, dialogue or voices in GTA 6. It also doesn't give us an inventory of every piece of software used to build it. Turning a broad discussion of technology into a specific production revelation would go further than the evidence.

The Word AI Covers Several Different Questions

A player asking whether GTA 6 uses AI could mean the behaviour of its computer-controlled characters, the tools used during development, or generated material appearing in the finished game. Those questions concern different systems and creative decisions. Zelnick's emphasis on the term's ambiguity helps explain why the discussion so often becomes muddled.

For developers, the distinction has practical consequences. A hypothetical tool that helps find a broken animation transition would need a different evaluation from a model producing dialogue for a character. The first might help a team inspect its own work. The second would affect the writing a player hears, raising questions about performance, consistency and authorship.

Neither example tells us what Rockstar has used. They illustrate why a useful answer needs the task and the tool attached to it. “AI was involved” is a thin description of a production process, particularly when the same label can cover systems with very different jobs.

Vice City Needs More Than a Supply of Assets

There is a positive case for AI here that doesn't require pretending a generated asset is a finished game. A tool could help a developer explore more possibilities, find faults earlier or spend less time on a repetitive task. Any benefit would depend on the team being able to check and use the result.

GTA 6 is a particularly good setting for thinking about that distinction. In an open-world game, a street layout affects movement and sightlines. A stretch of road can shape a chase. A conversation needs to fit the scene around it. Producing the individual pieces leaves a considerable amount of design work in deciding how they belong together.

Our view is that AI's strongest argument is helping people make those decisions with better information and more time to iterate. If a tool makes an experiment cheaper to try, a team can test an idea that might otherwise have stayed in a notebook. The creative value arrives when someone recognises what works and develops it further.

The Useful Measure Is What Reaches the Player

Zelnick's interest in efficiency is a statement of intent, rather than a measured result for GTA 6. The comments quoted this week don't establish a reduction in its budget or development schedule. They certainly don't support putting a number on how much time any unnamed tool has saved Rockstar.

For another studio considering AI, those measurements would be worth making on a manageable task. Count the time spent reviewing and repairing the output as well as generating it. Then ask whether the finished work serves the game. Faster production matters most when the team can keep the quality it wanted.

That's the part of Take-Two's argument worth following beyond launch. A commitment to authored worlds can coexist with curiosity about better tools. Developers should have room to investigate them, and players should get clear explanations when those choices affect the work they experience.

Sourcing note: The latest interview comments are drawn from Insider Gaming's 5 November reporting, with February context from GamesIndustry.biz and the release date from Rockstar's official site. This is analysis of public statements, rather than a hands-on report or an audit of Rockstar's tools. Featured screenshot: Rockstar Games.

Where would you most like AI tools to help a team building a world like GTA 6's? Let us know your thoughts in the comments. And if you've got a game you'd like us to review, send it over to editor@gamevibe.io.