Promethean AI promises to unlock experimentation in AAA development

There is an argument that AAA video games have become increasingly homogenised over the last few years, and studios have left their experimental inclination at the door in favour of a more commercially viable direction.

But when the cost of developing a single game frequently approaches $100 million, experimentation is less of an option with every single dollar spent.

As consumers baulk at the very notion of increasing game prices -- despite having remained largely static for 20 years -- what choice do studios have other than to move toward the middle ground; to cut loose the weird and wonderful, and let it float to the fringes of indie development?

This is just one of the many problems former Naughty Dog technical art director Andrew Maximov hopes to solve with artificial intelligence. Promethean AI, much like its mythological namesake, promises to enable progress -- but rather than stealing the gods' fire to uplift pre-civilisation man, Promethean AI aims to take care of the mundane work while leaving designers with room to create, rather than just produce.

Promethean is a standalone application that works most 3D editors. Artists simply put in a request, such as a nerdy, messy, '80s teenager bedroom as displayed in the demo video, and Promethean will build the space using developer created assets. It will learn based on what a given individual teaches it, and semantically aware and able to grasp abstract concepts such as messiness.

With Promethean, founder and CEO Maximov hopes to reclaim experimentation in an industry that is "very much over-generalised", and allow developers to explore long-neglected genres by reducing the time and cost of making games.

"We release a handful of really big products a year that everybody has to like," he tells GamesIndustry.biz. "That's the core idea, that every product has to cater to the widest audience possible in order for it to make economic sense. I think that once we bring down cost of production substantially, we'll actually be able to serve a lot more markets that are currently under-served."

According to Maximov one of the most exciting prospects of AI-assisted development is that "not every game will have to be an 80-hour behemoth that you need to figure out how to monetise".

In the face of rapidly escalating costs -- having increased nearly one hundred-fold since the PlayStation 1 era -- the AAA industry is looking towards business solutions like microtransactions and loot boxes, argues Maximov, rather than bringing down development costs.

"We're reaching a point in the foreseeable future where whatever we're doing right now is just not going to be sustainable"

"So instead of figuring how to make things cheaper, we figure out how to monetise our players more, which is both not something our players appreciate, but also not something we want to do as developers," he says.

"If we want to build a sustainable healthy industry that's inclusive for everyone and allows for creative experimentation and allows for more voices to be heard, we just need to bring the cost down considerably."

Promethean AI has been built to co-author virtual spaces under the creative oversight of developers and artists; early trials show an increased efficiency of around 50%, but Maximov believes productivity could be improved tenfold as the AI learns and improves.

"We need to do that in a way that's engaging and maintains creative ownership," he adds. "Because the other aspect of that is we can just go full Minecraft where everything is 100% procedural, but then that removes a lot of artistic core competence from productions... In our particular case, every artist is fundamentally a programmer of the tool that is going to improve their workflow and speed up production. All you need to do to automate that production is just build some of that art and build some of that core competence so that we can extrapolate and allow you to build more, faster."

Watch on YouTube

Of course, it's impossible to talk about automation without considering the impact it will have on jobs. Dating back to the industrial revolution over 200 years ago, automation has caused concern among workers, and it's no different in the digital age.

"That is definitely a big concern and is something that we take extremely seriously and think a lot about," says Maximov. "But the tricky part is... our alternatives at this point are not particularly great because the growth that we've seen in cost is not slowing down, and if you extrapolate that curve, by the end of the next console generation we'll go from $100 million to $200 million to develop a game.

"We're reaching a point in the foreseeable future where whatever we're doing right now is just not going to be sustainable. It's not one of those cases where we do nothing and things will just stay the same."

"Once we bring down cost of production substantially, we'll actually be able to serve a lot more markets that are currently under-served"

But once production costs reduce, Maximov argues, the industry will be able to make lots of games that are "extremely liked by smaller groups" rather than a few games that are "somewhat liked by everyone".

"That will unlock a lot of value and demand that the industry currently can't afford," he says. "It's almost like if we were the film industry and we could only afford to make superhero movies and then we suddenly discovered how to make dramas and comedies and every other genre cost effective as well. It just means a lot more quality, creative jobs. But also more opportunities for creatives to pursue their passion on their own economic terms."

It was time spent leading teams at Naughty Dog, which Maximov describes as the "most creatively fulfilling place" he ever worked at, that made him realise something had to change. The role, he says, was to take care of the team and he would frequently ask himself: "How the hell do we make it better, easier, going forward?"

"After we shipped Lost Legacy, we were just extrapolating to the next generation and the growth in complexity we were going to see, and I was just not seeing a way that we as an industry were going to tackle that," he says. "And that is scary because of all of the things like the creative ownership and making sure we don't balloon the team sizes but also making sure that everyone can have a healthy work-life balance when they're choosing this creative career."

Developer fatigue is a major problem Maximov hopes to solve with Promethean. Game development has always been a creative industry, but Maximov argues that "the scope of individual responsibilities for every developer keeps shrinking", and that devs are "less creatively engaged" in the face of such huge projects.

"No one has ever gone into game development because they wanted to do UV unwraps, or build collision meshes," he says. "Everyone who's in this industry got inspired because they were transported by some magical world somewhere or some story that captivated them, and that's what everyone wants to bring into the world with their work; people in this industry go to great sacrifices to entertain other people through that medium.

"It will be a fascinating process of reconnecting back with that initial inspiration that I think a lot of us have lost... [and] actually allow us to own a lot more of that creative vision."