
TOKYO — Artificial intelligence emerged as the central force reshaping both game development and the player experience at Tokyo Game Show 2026, Asia's largest gaming exhibition, even as a wide slate of new titles competed for attention.
Pareidolia, a new anime-style character game from Nexon Games (225570) unveiled on the 17th, features AI voice recognition that lets characters respond directly to a player's voice. Developed in-house by the company's IO Division, the technology runs entirely on the device rather than routing through a separate server.
Instead of tapping on-screen options, players can speak directly or call out a character's name — the game refers to them as Mates — to interact. Breaking from the one-way text menus of conventional role-playing games, the system aims to create the sense of talking face to face in real time. Nexon Games said it optimized the feature with mobile performance and heat generation in mind.
The technology is also seen as promising because it could extend interaction with characters beyond the screen. "Given the nature of on-device processing, we need to review further how far the feature can be expanded," an official at Nexon Games said. "For now we are testing it mainly for calling characters and basic conversation, and we plan to broaden its use based on the response to this TGS demonstration."

Competition was also fierce among business-to-business solutions aimed at overhauling the entire game production pipeline. During the two-day business session on the 17th and 18th, an AI Technology Pavilion spotlighted AI-based production environments. This year 14 companies — including Adobe, Thinking AI, Studio51, ZEAL and Mesh AI — set up booths and showcased a broad range of working development tools.
Exhibitors led with solutions promising sharp cuts in development time and cost. Thinking AI unveiled a system that uses data analysis to predict when and why dormant users are likely to return. ZEAL's Story AI analyzes text, video and social media reactions to assess the emotional arc and consistency of a scenario. It scores the completeness of a new writer's manuscript and then offers tailored advice, such as noting that a happy ending resonates more strongly in a debut work.

Gains in the speed of generating game assets were also striking. Kim Jun-hyeok, who oversees Korean and Japanese business at Mesh AI, which exhibited 3D model generation technology, said the tool produces a 2D image within seconds of a request for something like a futuristic radio tower and then immediately builds a 3D model from it. "A process that used to take about a month by hand is finished in just three minutes," he said.
Elsewhere, Heroz, known for its long-running AI development work, promoted case studies of game AI built with publishers along with AI-based work support services tailored to the gaming industry. Among indie studios, several stood out for using generative AI as a core gameplay element.

Japan's game industry is accelerating its adoption of AI amid unease over the gap with the United States and China, the two leading AI powers. Sticking to traditional hand-crafted methods, the thinking goes, risks falling behind global rivals in development speed, productivity and overall quality.
A survey by the Computer Entertainment Supplier's Association, which organizes TGS, found that 85.8% of local game developers use generative AI. Respondents cited greater work efficiency and productivity as the top benefit, followed by shorter development times and lower development and operating costs. Innovation in AI infrastructure is emerging as a decisive battleground in the global game market.








