House House discloses AI coding help in Big Walk amid awards dispute
Some festival jurors reject generative AI use in any capacity. The studio describes a narrow debugging role and calls its lack of an internal policy a failure.
Loading page…
Some festival jurors reject generative AI use in any capacity. The studio describes a narrow debugging role and calls its lack of an internal policy a failure.
Listen to this story
House House says it used chatbot-assisted research to resolve a rendering bug during Big Walk’s development, but rejects treating its disclosure as an endorsement of generative AI. The revelation has sharpened disagreement among jurors for the 2027 Independent Games Festival: some say they will not vote for games using generative AI, while the festival has announced no such ban. House House says it will create an AI-use policy for future projects, making both the studio’s next rules and the festival’s January finalist announcement worth watching.
House House estimates that less than 0.1% of Big Walk’s code was connected to the chatbot use and says AI had no other role.
For the reported bug, a chatbot suggested a fix after other avenues failed; developers substantially revised it into about 10 lines of URP code.
The studio said the bug caused performance to drop by 50% for about 20 seconds at sunrise and sunset.
House House used AI chatbots to help solve bugs late in Big Walk’s development, a disclosure that has divided game awards jurors. In a statement published by Aftermath on November 10, the studio estimated that well under 0.1% of the game’s code was connected to that use and said generative AI played no other role.
The disclosure comes after a group of jurors for the 2027 Independent Games Festival said they would not vote for games developed with generative AI in any capacity. They also urged the festival to adopt a policy barring games containing AI-generated work. That was a position taken by some jurors, not an announced festival-wide ban.
Big Walk was submitted to the festival, whose finalists are scheduled to be announced in January 2027. Aftermath’s Luke Plunkett reported that the game’s AI use contributed to intense disagreement among judges and jurors. The jurors’ initial statement did not name particular games.
House House described using large language models, or LLMs, primarily to research problems involving third-party code through chatbots. Some responses supplied code snippets that developers copied as starting points, then changed before incorporating them into the game.
The studio offered one detailed example involving Unity’s Universal Render Pipeline, the rendering software referred to as URP. It needed to stop rendering distant objects that cast shadows, based on their distance from the camera. The existing approach measured distance in the direction the camera faced, leaving shadow-casting objects behind the player being rendered at an infinite distance, House House said.
According to the studio, this made the game run 50% slower for around 20 seconds at sunrise and sunset. Developers researched the issue and inspected the rendering code without finding a solution. They also spoke directly with a Unity engineer, who asked colleagues and ultimately concluded that the change House House wanted was not possible.
The team then asked an LLM chatbot. After several dead ends, it suggested a possible fix. House House said the suggestion required extensive modification but ultimately worked, producing about 10 lines of code inside URP.
We have some concern that voluntary disclosures like this might contribute to the normalisation of this technology, or its treatment as “inevitable”.
House House, in its statement published by Aftermath
House House said it originally wrote a version of its disclosure because doing so was a suggested part of the game’s IGF submission. It made the statement public because its use of the tools had become a matter of public interest.
It had not made a disclosure on Steam. House House’s explanation was that the platform does not request disclosure of code or chatbot usage.
The studio explicitly rejected treating its statement as an endorsement of AI tools, however minimal their use. It said it supported people working to limit generative AI’s harms to games and the wider world, and apologized that Big Walk’s LLM use might not meet their standards.
House House also acknowledged a gap in its own rules: Big Walk was developed without an internal AI-use policy. It called that absence a failure and committed to working toward a more comprehensive policy for future development.
Loading discussion...
Join the conversation
Explain whether the kind of AI use should change the judgment.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.