Larian Studios CEO Swen Vincke has now shared more details about how the acclaimed RPG studio will be approaching the use of generative AI tools for developing its next game, Divinity. Speaking to fans in an AMA on Reddit, the director walked back a previous statement that Larian would use generative AI to help with references for concept art, but did say that Divinity could include gen-AI assets in the finished game as long as the studio is “100 percent sure about the origins of the training data and the consent of those who created the data.”
The gaming veteran landed in hot water last month over comments in an interview with Bloomberg about how Larian was experimenting with AI tools in the production of its next game, including generating references for concept art, producing PowerPoints, and other early production work that could help speed up the prototyping trial and error phase. At the same time, the CEO doubled down on the idea that no AI elements would appear in the finished game and stated that the studio wasn’t turning to the controversial tools to try and cut its headcount.
In today’s AMA, which was promised in response to the prior backlash, Vincke said that AI will no longer be used at all in the concept art phase, but did suggest the studio is looking at ways to use gen-AI tools trained exclusively on the company’s own material. Here’s his full statement:
So first off – there is not going to be any GenAI art in Divinity.
I know there’s been a lot of discussion about us using AI tools as part of concept art exploration. We already said this doesn’t mean the actual concept art is generated by AI but we understand it created confusion.
So, to ensure there is no room for doubt, we’ve decided to refrain from using genAI tools during concept art development.
That way there can be no discussion about the origin of the art.
Having said that, we continuously try to improve the speed with which we can try things out. The more iterations we can do, the better in general the gameplay is.
We think GenAI can help with this and so we’re trying things out across departments. Our hope is that it can aid us to refine ideas faster, leading to a more focused development cycle, less waste, and ultimately, a higher-quality game.
The important bit to note is that we will not generate “creative assets” that end up in a game without being 100% sure about the origins of the training data and the consent of those who created the data. If we use a GenAI model to create in-game assets, then it’ll be trained on data we own.
Larian’s new position addresses two of the chief criticisms of generative AI tools. The first is that they’re used by companies to undercut and replace human creatives. The second is that they produce content stolen from the millions of people whose work they were trained on. A set of gen AI tools trained entirely on Larian’s own proprietary game assets could potentially side-step that second critique, though it also suggests that Divinity could include at least some assets that were generated with the help of AI.
Developers have long used machine learning and procedural generation to help create pieces of a game that humans can then go and tweak as needed. It would be impossible to craft the beautiful cosmos of No Man’s Sky by hand. The bulk of its universe is generated from algorithms designed by the team at Hello Games. But as technology advances, algorithms have been able to produce more and more of the stuff that players might find in a finished game, leading many fans to balk at things like sloppy gen-AI loading screen art and even at the use of algorithms to generate spoken performances for NPCs, a subject of some controversy for Arc Raiders.
