House House, the developer behind the popular indie titles Untitled Goose Game and Big Walk, has confirmed that Large Language Models (LLMs) were used sparingly during the development of its latest game. The confirmation came via an updated FAQ page on the Big Walk website, following a controversy at the Independent Games Festival (IGF).

IGF Jury Statement Triggers Response

A joint statement from a contingent of 2027 IGF jurors announced they would decline to vote for any game utilizing generative AI in any capacity. While the game in question was not initially named, online discussions, particularly on BlueSky, strongly suggested that House House's Big Walk was the subject of this sentiment. The situation was further clarified when Bruno Diaz, lead designer for Fallen London, confirmed that IGF judges were made aware of Big Walk's LLM usage, as disclosure was reportedly required by IGF rules.

House House's Explanation

In response, House House updated its Big Walk FAQ to address the use of AI tools. The developer stated that LLMs were employed during the "final stretch" of development, primarily to assist with technical challenges and bug fixing. "In some cases, snippets from LLM responses were copied as a starting point to be modified into game code," the FAQ reads. House House estimates that "well under 0.1 percent of Big Walk’s code" was influenced by this usage, and that generative AI was not used in any other aspect of the game's creation.

The developer also noted that disclosure was not made on Steam because the platform "does not request disclosure of code or chatbot usage." The FAQ was updated as the matter became a public interest.

Developer's Stance on AI

House House concluded its statement by emphasizing that the confirmation is "not an endorsement of, or argument for, the use of AI tools, however minimal." The studio expressed its support for efforts aimed at mitigating the negative impacts of generative AI on the video game industry and the wider world. The developer also apologized that Big Walk's LLM usage might not meet the standards set by those seeking to limit AI's harm, acknowledging the concerns raised by the IGF jurors and the wider community.