The Creators of Baldur's Gate 3 Details Its Use of Generative AI for Next Divinity

The team behind popular role-playing games like Baldur's Gate 3 and Divinity: Original Sin has recently unveiled its next major project, generating immense anticipation within the industry. However, follow-up comments from the company's co-founder have introduced a new dimension to the discussion, focusing on the studio's philosophy toward generative artificial intelligence.

AI as a Creative Assistant, Not a Substitute

In a latest message, Swen Vincke outlined that the team is employing AI technology for specific supporting tasks. These encompass enhancing presentation materials, creating rough concept art, and drafting draft text.

Importantly, Vincke stressed that the shipping content in the game will be crafted entirely by human writers. "Our team is developing everything ourselves," he stated.

Our studio is continuously increasing our roster of writers and are actively putting together writing teams.

Since concept art is being particularly mentioned — we currently have twenty-three concept artists and have job openings for further talent.

All our efforts we do is additive and aimed at enabling creatives to spend more time on actual creation.

Any machine learning application used well is a boost to a creative team workflow, not a substitute for their talent.

Responding to Feedback and Defining the Path

The revelation of employing this technology initially sparked concern among a segment of the player base. In reaction, Vincke offered more clarification on public forums.

"At Larian, we employ machine learning to explore references, similar to we use search engines and physical media," he stated. "During the initial brainstorming phase we use it as a rough outline for layout which we then replace with original concept art."

He continued, "Larian brings on artists for their inherent skill, not for their capacity to execute what a AI generates."

Key Areas of AI Integration

Vincke had earlier detailed the studio's practical method to AI and ML, grouping its use into key functions:

• Automation of Tedious Tasks: Areas like polishing mocap data, audio processing, and pipeline-specific tasks like adapting animations for different models.
• Fast-Tracked Experimentation: Using technology to quickly build simple mock-ups of mechanics to test concepts prior to full production.
• Long-Term Aspirations: Investigating how AI could in the future facilitate new forms of player agency, specifically in simulating unforeseen permutations in a vast role-playing world.

He explicitly affirmed that key artistic domains — like writing — are are in no way departments where the company is cutting artistic involvement. On the contrary, Larian is actively hiring in these very positions.

"We are neither shipping a game with machine-made assets, nor looking at cutting teams to replace them with artificial intelligence," Vincke summarized.