Phoenix Wright: Ace Attorney Developer Surprised by Game’s Use in AI Testing

The iconic courtroom drama game, Phoenix Wright: Ace Attorney, has recently been repurposed as a testing ground for evaluating the reasoning abilities of various artificial intelligence models. This intriguing use of the game has garnered attention from both fans and industry experts alike.

According to a report by Automaton, researchers at Hao AI Lab utilized Capcom’s celebrated detective game to assess four distinct AI models. The objective was to gauge their capacity for logic and reasoning. As stated by AI researcher K. Ishi, the test evaluated how well these AI systems could identify inconsistencies in witness testimony, select relevant evidence, and effectively challenge discrepancies.

None of the AI models successfully completed the game, but there were notable performances: OpenAI and Google’s Gemini 2.5 Pro advanced to Chapter 4, while Anthropic’s Claude 3.7 only reached Chapter 2, and Meta’s Llama-4 Maverick struggled and did not progress past Chapter 1.

Response from Ace Attorney Developer

It’s a great idea to measure the true reasoning ability of an AI by having it play “Phoenix Wright: Ace Attorney”.This index uses Ace Attorney to evaluate the AI’s practical ability to “find contradictions in testimony, select appropriate evidence to support them, and refute them most effectively”.As a result, the best lawyer was o1 ↓ pic.twitter.com/L8hdWVPZRP

— K. Ishi@Industrial Applications of Generative AI (@K_Ishi_AI) April 16, 2025

Masakazu Sugimori, a developer and voice actor associated with the game, expressed his astonishment at the application of Phoenix Wright in AI testing. In a post on X, he remarked, “I was surprised to learn that the game I dedicated myself to 25 years ago is now being used in this manner, especially internationally. It’s fascinating to see how the AI models falter at the first episode.”

Regarding the role of artificial intelligence, Sugimori maintains a positive outlook, believing that there is potential for humans and machines to coexist harmoniously. He stated, “The ways in which games can contribute to this field are intriguing. I am confident that humans will maintain a competitive edge over AI. While individual tasks may be achieved more efficiently by machines, the essential human role remains to think critically.”

Sugimori’s emphasis on the critical thinking capability of humans aligns with the observed difficulties the AI models faced during their challenge in Phoenix Wright. Therefore, if players find themselves struggling to navigate a particularly challenging court case, they may be better off relying on their own reasoning skills—or perhaps resorting to the classic strategy of “save scumming” for assistance!

Source & Images