Embark Studios has pulled off a truly impressive feat – making an extraction shooter appealing to the masses. 24 hours out from launch, Arc Raiders sits proudly as Steam’s third most-wishlisted game behind Subnautica 2 and Valve’s very own Deadlock. Aside from its more accessible approach to the extraction genre, which makes account wipes optional and gives players a path back to the good times even if they lose every piece of gear they’ve collected, I think that the game’s visual aesthetic and sound design have played a huge role in those wishlists. Yes, it’s set in a gritty, post-disaster world like so many other extraction shooters, but Embark’s flashy take on make-shift raider gear, imposing robot designs, and synthy score makes it really stand out from the crowd. However, many have noticed its vague artificial intelligence disclosure on Steam, so did AI play a role in Arc Raider’s visual splendor?
“Arc Raiders in no way uses generative AI whatsoever,” assures design director Virgil Watkins during my interview with him ahead of the game’s launch. “We use something called machine learning, or reinforcement learning, and that’s to do with the locomotion for our larger drones with multiple legs, but there’s no generative content whatsoever.”
I press on if that’s the only reason for the AI disclosure on Steam, citing how Embark used a text-to-speech system to create new voice lines in its first multiplayer game, The Finals. Watkins says he’s not totally sure, as he’s not responsible for the storefront page. However, he does confirm that the same voice line tech is present in Arc Raiders.
“As [you] stated, [it’s] the same as The Finals, we use that text-to-speech model. That is, we hire and contract voice actors for it – it’s part of their contract that we use it [AI] for this purpose, and that allows us to do things like our ping system, where it’s capable of saying every single item name, every single location name, and compass directions. That’s how we can get that without needing to have someone come in every time we create a new item for the game.”
Interestingly, in an interview with The Game Business earlier this week, the CEO of Embark, Patrick Soderlund, brought up the studio’s use of artificial intelligence. Talking about the original vision of Arc Raiders (which was a solely PvE experience before it was repurposed into the PvPvE game it is today), he says “the compulsion loop of the game was so dependent on deep innovations within artificial intelligence both in reinforcement learning, machine learning and other aspects. And we thought that we could build a whole game around that experience.”
Later in the interview, he talks about how Embark has used artificial intelligence and procedural generation to speed up its development process, including in “content creation pipelines.”
“If you ask for [development that’s] 100 times faster, you have to take what you know and basically throw that away and then start really thinking about what type of modifications and transformation do we need to go through in order to completely change the way that we approach this,” Soderlund says. “So that’s when we came into procedurally generated content, to using AI and machine learning to some extent in even the content creation pipelines, to what types of tools we build, and realizing that a lot of these tools that we’ve been working with for so long stem from software that’s 25 to 30 years old.”
So, AI has definitely been leveraged in Arc Raiders’ development, but Watkins is clear that generative AI hasn’t contributed to its striking visuals. You can read more of my chat with Watkins here, where we touch on the team’s Dark Souls-style approach to difficulty, what you’ll actually get for wiping, and much more.
Ahead of the game’s imminent launch, make sure you check out the Arc Raiders system requirements before diving in. Our guides on Arc Raiders Hatch Keys and the best Arc Raiders builds will also come in very handy.
Where do you stand on the use of generative AI in game development? Let us know in the PCGamesN Discord server.
