Earlier today Bloomberg’s Jason Schreier published an interview with Swen Vincke, the CEO of Larian, the company behind the internet’s favourite video game and demonic relationship simulator Baldur’s Gate 3. It could not have gone worse for the guy.
Among all the expected pre-release interview talk about how their next game, Divinity, will be making all kinds of improvements over their last game […]
Saying you only use AI for “reference” is wild. Artists use image searches and books as inspiration because they are drawing on art (but also everything else from colour palettes to photos to the weather). There’s experience there, things they can relate to, be inspired by. There is no inspiration in slop! Everything AI is presenting to you is simply stolen and amalgamated. It’s like asking your phone’s autocorrect for relationship advice.
There’s an unsurprising tendency across leaders in the tech world (games included) to see art as merely part of something’s production line, a box that needs to be ticked before copies can be sold. It’s why AI is often justified as something that saves time, or saves money. But with art, that process is the point. The themes and ideas artists draw on, the way they iterate through those ideas with sketches, the work itself is what creates art. There are no shortcuts.
Under Vincke, Larian has been pushing hard on generative AI, although the CEO says the technology hasn’t led to big gains in efficiency. He says there won’t be any AI-generated content in Divinity — “everything is human actors; we’re writing everything ourselves” — but the creators often use AI tools to explore ideas, flesh out PowerPoint presentations, develop concept art and write placeholder text.
The use of generative AI has led to some pushback at Larian, “but I think at this point everyone at the company is more or less OK with the way we’re using it,” Vincke said.
This reads like “some part of his
salarystock portfolio depends on it” copium.Holy fuck guys we’re not “pushing hard” for or replacing concept artists with AI.
We have a team of 72 artists of which 23 are concept artists and we are hiring more. The art they create is original and I’m very proud of what they do.
I was asked explicitly about concept art and our use of Gen AI. I answered that we use it to explore things. I didn’t say we use it to develop concept art. The artists do that. And they are indeed world class artists.
I think my, and maybe others, main issues are:
- I don’t know the training data of the AI tools
- I don’t know the actual use in the workflow
If the training data is the same as the popular image gen ones, then regardless of how it’s used I don’t support it. If a company is charging for/building a brand on AI tools the artists whose work trained the tools need to be compensated. If they are working with or paying for any of these models I think it supports a business that is stealing from artists.
If you don’t know what the training data is, there’s no way to know for sure it’s not replicating someone else’s work in whole or in part in a way that an artist would not. Artists know how to take inspiration and pay homage without stealing, but if they don’t actually know the original work that the AI is basing things off of, there’s no way to ensure that.
To me it sounds like they’re using generative AI images for inspiration boards, which typically is made from other people’s art. If this is the case, then there’s no more risk of stealing art on accident than usual (I hope they have better practices than Bungie).
I used gen ai for some concept stuff, not as a finished product, but as a muse. I haven’t touched it since the first generation of image gen, so to say what it made for me was nowhere near what I had in mind is a dramatic understatement (boy did it suck with fins!). But it pointed me in the direction I wanted and gave me some additional ideas for what to use for my own work, as well as posture models (I don’t really have art skills as such; that trait largely skipped me in my wildly artistic family, but I’m trying to develop it somewhat cuz I have IDEAS). It helped fuel my idea spark into a creative bonfire.
If I showed you the finished piece (which isn’t done by a long shot, and probably won’t be as good as I imagine it), then the muse concept stuff, you’d be hard pressed to find any real resemblance, other than certain major elements being present, but those elements were what I was asking it to generate in the first place.
I assume that is basically what their art department is doing as well. And for that, it’s actually pretty useful, specifically because it pulls details from many different works, and blends them in potentially novel ways.
What does he think concept art is for?
Feels like people are assuming they’re generating images and calling it concept art but there are so many ways an artist could use it in the process of making concept art. A process that starts with pumping out as many minimal-effort concepts as possible… Do you really want to colorize 20 sketches that will likely get scrapped or would you rather stick it in a homebrewed ComfyUI workflow and have a slightly worse version done in 30 seconds?
I know this is the fuck ai community and I do agree with the sentiment but hating on “AI” as a vague idea is a bit ignorant. There are legitimate uses for it, they’re just overshadowed by the awful ones getting pushed to us by corporations.
I’m as big of an AI hater as anyone, but if you’re going to use it, this is pretty much the way you do so.
None of the results end up in the final product, concept art is notoriously different from final results in almost all projects, and having a concept artist spit out 30 basic AI concepts in a day and then pick 2 or 3 of them to manually iterate further on is, actually, streamlining their work flow without major upset and without completely replacing the artist in question.
Would I rather they aren’t using Gen-AI at all? Yeah, absolutely, I think it is one of the great evils of the modern world. But if you do insist on using it within the creative process, this is one of the least upsetting and most sensible ways I can think of doing so. I don’t see this as problematic beyond the fact that Larian is supporting AI companies by doing so.
having a concept artist spit out 30 basic AI concepts in a day and then pick 2 or 3 of them to manually iterate further on
My interpretation of what he said is that the concept artists are using AI concepts to fill out their mood board.
I fucking hate AI, but this is a stupid fucking take on why its bad. Guess we gotta go back to hand drawn animation, computer animation takes away from the process. Guess we gotta hand paint family photos cause cameras take away from the process. As long as workers are keeping their jobs, the environment isn’t being destroyed for it, and the end result was ultimately made by a human… I don’t fucking care. And in this case, we are 2/3. So the only problem to me here is the environmental impact, and they could easily fix that through self-hosting the AI model.
Using is for placeholder text is probably overkill but doesn’t seem much different than ipsum lorem which has been used for years.
Lorem ipsum is at least easy to spot, the trouble with using AI is that it’s believable enough to get overlooked. This is what happened to The Alters and Claire Obscur.
Purity test bullshit. AI is brilliant at helping you pull an idea from your brain to create a rough draft. Why are so many people’s thoughts processes becoming as black-and-white as conservatives’?
