In its recent shareholder briefings, Capcom has said that it will not be using AI-generated assets in its games, but that it does plan to use the technology to boost its efficiency and productivity in game development. While these statements have been quite general, Capcom’s engineers recently shared some more details on their policy regarding generative AI, as well as the nature of the company’s collaboration with Google Cloud and agentic AI adoption.
Talking to 4Gamer, Capcom’s game development platform and AI solutions VP Shinichi Inoue says the company does not plan to use AI for the creation of graphical and other game assets. “What we in the entertainment industry consider extremely important in contrast to artificial intelligence is human sensibility. Even top-tier AI still cannot match our creators when it comes to sensibility. That’s the current reality. Because of that, concentrating human effort on work requiring this kind of sensibility is more efficient from the standpoint of human capital management, and it’s also important for coexistence with creators,” he explains.
Interestingly, this seems to somewhat contradict Capcom’s experimental AI initiatives from 2025, when the company announced it was working with Google Cloud to figure out a way to streamline the process of brainstorming and conceptualizing the tens of thousands of objects needed for an AAA game (an example used at the time was that of coming up with a fictional TV design for a game’s environment). While this workflow was intended for the proposal stage and not final asset production, it seems to clash with Capcom’s current stance on not using AI for the creation of in-game assets, suggesting the company might have changed its views somewhere along the way.
What is Capcom using AI for, then? According to Inoue, streamlining testing and communication-related tasks is the most urgent challenge AAA development faces, because this “routine work” increases exponentially as games grow in scale. That’s why his team is working on a playtesting system that leverages Gemini alongside several AI agents trained in-house.
“The AI reports its findings to debugging check agents, but rather than leaving everything to humans, another agent first evaluates the reports against the game director’s concept. A huge volume of this checking and evaluation work is carried out while humans are asleep. After that, the system screens for issues that are highly likely to be ‘incorrect when compared to the game’s intended concept’ and presents them. Because the AI performs these checks in advance, humans no longer need to carry out enormous amounts of manual verification work themselves,” Inoue tells 4Gamer.
According to a press release by Google Cloud from April, Capcom’s playtesting system handles about 30,000 hours of testing work per month, freeing up developers’ hands significantly. The AI agents use Gemini Vision to “look” at the game screen and discover technical failures, while distinguishing actual errors from intentional design choices based on data they’re trained on. Additionally, the AI analyzes past data to predict where the game system is likely to “break” next and prioritizes testing high-risk areas instead of testing randomly.
