The Creators of Baldur's Gate 3 Details Its Use of Machine Learning for Next Divinity

The developer behind acclaimed role-playing games like Baldur's Gate 3 and Divinity: Original Sin has recently teased its next major project, sparking immense excitement within the player base. However, recent comments from the company's lead designer have introduced nuance to the narrative, touching on the studio's approach toward generative artificial intelligence.

AI as a Creative Assistant, Not a Substitute

In a new clarification, Swen Vincke outlined that the team is utilizing generative AI for specific preliminary functions. These encompass fleshing out PowerPoint slides, producing early-stage artistic references, and drafting temporary copy.

Importantly, Vincke emphasized that the final content in the game will be created entirely by human writers. "We are creating all the content in-house," he stated.

We are actively increasing our team of writers and are actively forming writing teams.

Given that concept art is being particularly referenced — we right now have twenty-three artistic staff and have positions available for further artists.

All our efforts we do is supplementary and aimed at having people spend more time on actual creation.

Any machine learning application applied correctly is supplementary to a developer's process, not a replacement for their craft.

Responding to Feedback and Defining the Path

The news of AI usage initially sparked concern among a segment of the community. In reply, Vincke provided additional detail on online platforms.

"At Larian, we employ AI tools to gather inspiration, just like we use the internet and art books," he wrote. "In the very early brainstorming phase we use it as a rough outline for layout which we then substitute with authentic artwork."

He added, "Larian brings on creatives for their creative vision, not for their capacity to execute what a algorithm proposes."

Three Pillars of Practical Application

Vincke had in the past broken down the studio's practical strategy to machine learning, defining its use into primary functions:

• Handling Monotonous Jobs: This encompasses polishing mocap data, voice editing, and pipeline-specific tasks like retargeting animations.
• Fast-Tracked Experimentation: Using systems to quickly build basic versions of gameplay ideas to experiment with concepts before expensive implementation.
• Future Potential for Gameplay: Investigating how AI could one day facilitate new forms of player agency, particularly in creating player-driven narratives in a vast role-playing world.

He explicitly noted that central narrative areas — including visual art — are not fields where the company is reducing artistic input. Conversely, Larian is recruiting more in these very fields.

"We are neither launching a game with machine-made assets, nor looking at reducing creatives to replace them with artificial intelligence," Vincke stated definitively.