AI is 'a combination of metadata with a parlor trick,' Take-Two boss says_ 'A great thing' for busin

Take-Two chairman and CEO Strauss Zelnick has never been shy about sharing his thoughts on artificial intelligence, and he's been pretty consistent about it over the years, too. While some game industry CEOs see the rise of AI as an imminent revolution in revenue-generation (and yeah, sure, creativity and whatnot), Zelnick has always been a bit cooler on it: "," he said in May 2023, "and I believe will stay that way."

And so he was today, speaking at the Paley International Council Summit in Menlo Park, California, where he said that [[link]] the use of AI will "probably be able to create a bunch of efficiency" for game studios, but won't produce any hit games because it can't actually be creative.

"What AI is, [[link]] is the combination of big datasets with a bunch of compute within natural LLM—a large language model," Zelnick told his audience. Leaning into an opinion he shared , he continued, "And by definition, a data set is what? Backward [[link]] looking. By definition, creativity is what? Forward looking. And to the extent that AI appears to be forward looking, it is what? A predictive model."

This, he continued, is where the wheels come off. Predictive models do their jobs very well when they have enough clear data, but are far less capable when they don't have as much to work with. AI seems so extraordinary now because it's "a combination of metadata with a parlor trick," a description I think really cuts to the heart of the thing, but it will become mundane in time—"we're all going to get used to it," as Zelnick put it—much like we did with Google.

Zelnick also predicted, , that the advent of AI will not reduce employment, but will in fact increase it: "In 1865, 65% of the US workforce was involved in agriculture. Today we make food for America and food for the rest of the world. And 2% of the workforce is involved with agriculture. And I defy you to find anyone who recently said to you, 'It's so horrible, I can't get a job as a farmer.'"

(He used that exact same analogy in those prior comments about the impact of AI on employment, too. The man really is consistent.)

Despite his apparent doubts, Zelnick is not an AI naysayer—he just seems to more readily perceive its limitations than other game industry executives, or at least be more willing to call them out. "AI is a great thing," Zelnick said. "It's a great thing for every industry. Will it recreate or create genius? No. Will it create hits? No. It's a bunch of data with a bunch of compute with a language model attached."