The intense focus on generative AI and large language models (LLMs) within the game development industry is polarizing, according to Dr. Luke Dicken, the former head of AI at Take-Two. Players are increasingly scrutinizing game development tools, with many checking Steam store pages for mentions of AI usage.
Take-Two, the parent company of Rockstar Games, recently dissolved its dedicated AI research division, a move that surprised many given the industry's broader embrace of AI tools. Notably, most of the employees laid off from this team were not directly involved with generative AI. The team originated at mobile developer Zynga, which Take-Two acquired in 2022, and its scope later expanded to encompass all of Take-Two.
Dr. Dicken, speaking with GamesIndustry.biz, revealed that his team was established in 2019, predating the public launch of ChatGPT and the subsequent surge in generative AI interest. He expressed that generative AI was not his primary passion. "It was something I think there's a moral obligation to see managed as best as can be," he stated, adding that by 2025 or 2026, a complete absence of generative AI in a large corporation's toolkit would likely draw criticism. He emphasized the need to push back against the more extreme applications of generative AI, acknowledging its controversial nature.
Dicken's true interest lies in a more comprehensive approach to AI in game development. He recalled instances from five years prior where a novel algorithm for accelerating level generation in mobile titles was met with skepticism. "Back then, people looked at us like we had two heads," he said. Now, however, the prevailing AI hype has made stakeholders more receptive, to the point where even outlandish claims about AI's capabilities are met with acceptance.
This heightened receptiveness, Dicken noted, has made people more open to concepts that were previously dismissed, such as the potential of traditional AI techniques. However, he fears that the current overblown excitement surrounding generative AI could ultimately harm the entire field. "My worry is that generative AI is poisoning the well," Dicken stated. He believes there is insufficient nuance in the current discourse to preserve the value of traditional AI methods, suggesting that the industry has already entered a entered a 'trough of disillusionment' with LLMs.