Larian CEO Explains AI’s Nuanced Role in New Divinity Game
▼ Summary
– Larian Studios is using generative AI for early-stage tasks like concept art outlines and presentation materials, but final game content will be created by human artists and writers.
– The studio’s co-founder clarified they are not replacing creative staff with AI, but are hiring more concept artists and writers while using AI as a tool to improve workflow efficiency.
– Larian employs AI and machine learning primarily for automating non-creative tasks, rapid prototyping (“white boxing”), and testing narrative consistency, not for core creative work like writing or final art.
– Fan reaction to the initial report was negative, prompting detailed clarifications that the studio’s goal is to use AI to make creative work easier, not to reduce team sizes.
– The studio’s position is that AI tools are additive to the creative process, aimed at accelerating iteration and handling complexity, while human talent remains central to game development.
The recent announcement of a new Divinity game from Larian Studios has ignited passionate discussions, particularly regarding the studio’s approach to generative AI. Following the reveal, co-founder Swen Vincke provided detailed clarification to address concerns, emphasizing that AI serves as a supportive tool for specific, non-creative tasks, not as a replacement for human talent. The studio is actively expanding its creative teams, including concept artists and writers, while leveraging machine learning to handle technical and repetitive processes, ensuring the final game’s narrative and artistic vision remain firmly in human hands.
Vincke explained to Bloomberg that the technology is used for practical, behind-the-scenes work like fleshing out presentation slides, generating early concept art for composition reference, and creating placeholder text. He was adamant that everything appearing in the final game will be crafted by people. “We’re writing everything ourselves,” he stated. This distinction, however, did not prevent a strong reaction from some fans concerned about the integration of AI in game development.
In response to the feedback, Vincke provided a comprehensive statement to Polygon. He outlined Larian’s ongoing investment in human creativity, noting increased hiring for concept artists, writers, and translators, and the formation of dedicated writer rooms. He framed the use of machine learning as an additive tool aimed at improving workflow. “Any ML tool used well is additive to a creative team or individual’s workflow, not a replacement for their skill or craft,” Vincke wrote. He firmly denied plans to release a game with AI-generated core content or to reduce team sizes, stating the goal is to research tools that make the daily work of developers easier and more efficient.
Vincke took to social media to further clarify, strongly refuting a characterization that Larian was “pushing hard” to replace artists. “Holy fuck guys we’re not ‘pushing hard’ for or replacing concept artists with AI,” he posted. He compared the use of AI in early ideation to using Google or art books for reference, a starting point that is always replaced with original artwork. “We’ve hired creatives for their talent, not for their ability to do what a machine suggests,” he added.
This philosophy is consistent with Vincke’s earlier comments to GameSpot, where he detailed three primary uses for AI and machine learning at Larian. The first is automating tedious tasks like cleaning motion capture data or voice editing. The second is “white boxing,” which accelerates prototyping by allowing designers to test gameplay ideas more rapidly before full implementation. The third, and most experimental area, is exploring new gameplay possibilities, particularly for creating deeper reactivity to player choices in an RPG, though Vincke admitted this application is not yet mature.
Crucially, Vincke draws a line at using AI for core creative expression. He emphasized that areas like writing, visual art, and music are seeing increased human investment, not divestment. Instead, machine learning is targeted at managing complexity, such as through “narrative validation”, an automated process to help check for inconsistencies across hundreds of story permutations, a task that is vital but not creatively fulfilling.
Whether this nuanced explanation will fully reassure all fans is an open question. However, Larian’s position is clear: AI is a tool for iteration and efficiency, while the soul of the game, its story, characters, and world, will be unmistakably human-made. For players, the only “slop” they might encounter will be the intentionally disgusting kind found in the game’s world.
(Source: Polygon)
