Game developer Tomo Kihara offers a new path

By Edward Brydalski
(Image above: Deviation Game by Tomo Kihara & Playfool — Daniel Coppen & Saki Maruyama. Photo by Aya Kawachi at Miraikan, the National Museum of Emerging Science and Innovation in Tokyo)

Deviation Game by Tomo Kihara & Playfool (Saki Maruyama & Daniel Coppen & Saki Maruyama)

Remember when AI was just another word for bots¹ in a video game?

Twenty-one years ago, many praised F.E.A.R.’s enemy soldier AI as it responded to the player’s tactics and changed its own on the fly. 

Just a couple of years later, Akinator showed us what machine learning² was capable of:  remembering millions of users’ inputs to give the illusion that it could read the player’s mind. Good AI was synonymous with games that could make players feel like their actions made a real impact in the games they play.

In 2026, AI is on the verge of making “games” of its own. Many people feel skeptical, believing that game studios and the firms that fund them have taken the wrong lesson to heart. The Google Trends for searches of “AI slop”³ is the canary in this coal mine.

That leaves us with a fatal question: Is there any hope for a games industry married to AI technology?

Tomo Kihara thinks there is hope. Conditional hope.

Kihara is an artist and game designer based in Tokyo. He spent time in Rochester earlier this year, including a panel appearance at the Strong National Museum of Play. His bread and butter is making games that use AI to enhance and encourage human creativity, rather than supplant it.

Case in point: Deviation Game.

Kihara started creating Deviation Game in 2022 alongside art and design duo Playfool (Daniel Coppen and Saki Maruyama). It was a featured attraction in several festivals and galleries across Asia and Europe; finalized in 2025, it is available for purchase right now on Steam.

The basis is to draw pictures based on prompts, then your fellow players have to guess what your prompt was. But here is the unique part: An AI is guessing at your prompt as well. The real goal is for your picture to draw your prompt in such a way that your fellow humans can guess it, but the AI can’t.

“It draws parallel to Alan Turing’s Imitation Game,” Kihara elaborated in a Zoom interview in March of 2026. You may know it as the Turing Test⁴

“It asks whether these like systems can copy us,” Kihara said, “and reverses the premise in a sense that, can we use it to deviate from these machines?”

Whereas many games use generative AI to create art for games, Deviation Game uses AI as a perceiver, a creative modifier to grow a greater breadth of creativity in the player.

“I would use the word Game Master⁵,” Kihara explained, “So the AI is not in the center of the game, but is used as a moderator to push human creativity forward.”

But unfortunately, in 2026, many games do use generative AI to create game assets. And time and again it brings about a serious ethical dilemma: Companies that use generative AI models to create art do not pay or give credit to the human artists whose works trained the AI.

When you prompt Google Gemini to make a “shiba inu dog riding a skateboard in a watercolor art style,” that model isn’t making original decisions on brush strokes, color palettes or use of negative space. It’s regurgitating the decisions made by the human creators of the millions of artworks the AI has scrubbed, logged and categorized. Hell, the prompter can even ask an AI to emulate an individual artist’s trademark methods to a shockingly accurate degree.

It’s a legal gray area that the industry hasn’t adopted a solution for yet. Thankfully, Kihara might have one.

The Game of Possible Lives was unveiled at Tokyo’s Mori Art Museum in 2025. It is an AI-driven life simulation game inspired by Alter Ego (1986) and the original Checkered Game of Life (1860). The game uses a locally-hosted large language model⁷ to generate a statistically plausible person for players to inhabit, guiding them through major choices and consequences until their death.

The game’s pixel art is mostly generated by an AI model. But instead of using a model that scrapes the internet for whatever images it can find, Kihara took a more ethical approach. For image generation, he used an AI model that was trained on artwork by artists that gave consent and were compensated. 

“Let’s say, there’s an artist that you want to work with and a model that they made,” Kihara explained, “Let’s say every time the image gets generated, there’s a percentage fee that’s paid to the artist. There could be a fairer way for this to be done.”

Not only does this mean the artist actually gets paid to be copied by an AI, but it allows the developer to generate a wide range of images at a moment’s notice and in a consistent, purpose-built art style. 

“(AI) is a changemaker,” Kihara concluded, “It can make connections, but it also makes conflicts. We just have to be aware of what kinds of changes it is making.”

The games industry would be better off by putting Kihara’s ethical approach on AI into practice. And we would all be better off by learning to discern games that use AI to improve the lives of players and artists, not exploit them.