Machine-learning and artificial intelligence systems are some of the most-talked-about technologies in gaming--and across a variety of other industries. Some fear advancements to such systems might lead to negative consequences, while others predict a rosier future. Still others, like Larian Studios boss Swen Vincke, are trying to thoughtfully implement new technologies to help empower developers, not replace them. We caught up with Vincke recently, who discussed a range of subjects, including machine learning and AI. Watch the full interview below.
Vincke said Larian currently uses machine learning systems to help automate game-development tasks that "nobody wants to do."
"It's the obvious things, like a motion capture cleaning or voice editing or, something very specific. Retargeting. So that is basically if you play with different species, you want to be able to reuse an animation on a different species that are in different size, and then they're doing certain interactions with others," Vincke explained.
For these purposes, machine learning "works really well," Vincke said. Larian is also embracing the technology for what's called "white boxing," which is the term for how a studio like Larian iterates on ideas rapidly. "White boxing is essentially the stage before you actually do the real implementation," he said, noting that using machine learning can help speed up the process.
Under traditional development circumstances, this might take "some time" before a developer can create something that's actually playable, but white boxing with machine learning helps "accelerate that."
But for Vincke's money, the most exciting implementation of machine learning is when it can be used for new gameplay elements. However, it's still early days, and Larian is "not there yet," or even close. But it's an area that Vincke is excited about because it could help the studio implement new RPG ideas faster.
"For an RPG developer, what you really want is something that helps with reactivity to agency," he said. "So, permutations that you did not foresee, reactions to things that the player has done, in the world, they will certainly augment the gameplay experience."
Vincke mentioned how in a game like FIFA (now called EA Sports FC), the under-the-hood procedural generation happens in real time, and that's what he's referring to in regards to how machine learning can impact gameplay.
For all of this hope and optimism about using new technologies like machine learning to help speed up and improve game development, Vincke said he understands there are "red flags" to consider. Some people have legitimate fears of machine learning and AI systems stealing their work or taking their jobs. However, Vincke said Larian is hiring more people, not fewer, thanks to new technologies. This is what others, like EA CEO Andrew Wilson and Take-Two boss Strauss Zelnick, have predicted.
"It's always going to be the people that are going to make a difference," Vincke said. "And so for us, it's super clear that we're investing heavily into the team and the human touch that's being added to that."
With more automation, human workers are freed up to create more, which in turn leads to the possibility of making "more complex things," Vincke said.
"When complexity goes up, things start getting really interesting. So it's fair to say that the games that are being made today are arguably more complex than they were in the past," he said. "The faster you can iterate, the faster you will get to a certain result. And then and the tightening of those pipelines should allow you, in theory, to make your games in a more efficient manner."
Automation with machine learning AI is a "huge thing," Vincke said. People may not be aware of how much time an animator may spend on "stuff they don't want to work on." As such, thanks to automation, "It adds a lot to the time that they have to work on the creative stuff," Vincke said.
Vincke said it's "the wrong attitude" to assume that automation will replace various people and departments. Because automation should, in theory, allow people to do "more of what really matters and less of the stuff that they don't want to do."
"That's not necessarily cost reduction," he added. "But it's going to be an increase of ability, which is a good thing."
Following the huge success of Baludr's Gate 3 in 2024, everyone is wondering what Larian is going to make next. We know it won't be Baldur's Gate 4, but instead a project codenamed Excalibur.
Beyond Larian's next game, Vincke is thinking even further out and has grand plans to create a "very big RPG" that will "dwarf" any of Larian's previous releases. "It's an inspirational goal. That's it. I cannot say more, because it needs to stay in that dream-reality," he said.
While Larian will not make Baldur's Gate 4, Hasbro has said it is keen to expand the Baldur's Gate series with other teams. "We're going to take our time and find the right partner, the right approach, and the right product," Hasbro previously said about the future of Baldur's Gate.
Source
Tasks "Nobody Wants To Do"
Vincke said Larian currently uses machine learning systems to help automate game-development tasks that "nobody wants to do."
"It's the obvious things, like a motion capture cleaning or voice editing or, something very specific. Retargeting. So that is basically if you play with different species, you want to be able to reuse an animation on a different species that are in different size, and then they're doing certain interactions with others," Vincke explained.
For these purposes, machine learning "works really well," Vincke said. Larian is also embracing the technology for what's called "white boxing," which is the term for how a studio like Larian iterates on ideas rapidly. "White boxing is essentially the stage before you actually do the real implementation," he said, noting that using machine learning can help speed up the process.
Under traditional development circumstances, this might take "some time" before a developer can create something that's actually playable, but white boxing with machine learning helps "accelerate that."
But for Vincke's money, the most exciting implementation of machine learning is when it can be used for new gameplay elements. However, it's still early days, and Larian is "not there yet," or even close. But it's an area that Vincke is excited about because it could help the studio implement new RPG ideas faster.
"For an RPG developer, what you really want is something that helps with reactivity to agency," he said. "So, permutations that you did not foresee, reactions to things that the player has done, in the world, they will certainly augment the gameplay experience."
Vincke mentioned how in a game like FIFA (now called EA Sports FC), the under-the-hood procedural generation happens in real time, and that's what he's referring to in regards to how machine learning can impact gameplay.
Red Flags
For all of this hope and optimism about using new technologies like machine learning to help speed up and improve game development, Vincke said he understands there are "red flags" to consider. Some people have legitimate fears of machine learning and AI systems stealing their work or taking their jobs. However, Vincke said Larian is hiring more people, not fewer, thanks to new technologies. This is what others, like EA CEO Andrew Wilson and Take-Two boss Strauss Zelnick, have predicted.
"It's always going to be the people that are going to make a difference," Vincke said. "And so for us, it's super clear that we're investing heavily into the team and the human touch that's being added to that."
With more automation, human workers are freed up to create more, which in turn leads to the possibility of making "more complex things," Vincke said.
"When complexity goes up, things start getting really interesting. So it's fair to say that the games that are being made today are arguably more complex than they were in the past," he said. "The faster you can iterate, the faster you will get to a certain result. And then and the tightening of those pipelines should allow you, in theory, to make your games in a more efficient manner."
More Time To Be Creative
Automation with machine learning AI is a "huge thing," Vincke said. People may not be aware of how much time an animator may spend on "stuff they don't want to work on." As such, thanks to automation, "It adds a lot to the time that they have to work on the creative stuff," Vincke said.
Vincke said it's "the wrong attitude" to assume that automation will replace various people and departments. Because automation should, in theory, allow people to do "more of what really matters and less of the stuff that they don't want to do."
"That's not necessarily cost reduction," he added. "But it's going to be an increase of ability, which is a good thing."
What's Next For Larian
Following the huge success of Baludr's Gate 3 in 2024, everyone is wondering what Larian is going to make next. We know it won't be Baldur's Gate 4, but instead a project codenamed Excalibur.
Beyond Larian's next game, Vincke is thinking even further out and has grand plans to create a "very big RPG" that will "dwarf" any of Larian's previous releases. "It's an inspirational goal. That's it. I cannot say more, because it needs to stay in that dream-reality," he said.
While Larian will not make Baldur's Gate 4, Hasbro has said it is keen to expand the Baldur's Gate series with other teams. "We're going to take our time and find the right partner, the right approach, and the right product," Hasbro previously said about the future of Baldur's Gate.
Source