Portal writer Erik Wolpaw is "not worried about AI taking over creative writing," but it could be the straight-faced foil to GTA's chaos

Valve writer Erik Wolpaw, one of the leading scribes behind Portal and a long-time staple of the developer’s games, has revealed the ways he and several co-workers have been experimenting with generative-AI tools and what they’re capable of. Speaking on the MinnMax podcast, Wolpaw says he believes there’s potential for its use in reactive dialog, or to play the “straight man” to high chaos games like Grand Theft Auto. However, he doesn’t think it’s there yet, and notes he’s “fully against” AI generation being used as a developmental cost-cutting measure.

After making his name working on Psychonauts, Wolpaw joined Valve and worked alongside fellow writer Chet Faliszek on the Half-Life 2 episodes and TF2, before bringing us the masterworks of Portal and its sequel. While he left the company in 2017, he then returned two years later and has since contributed to the likes of Half-Life: Alyx. When the topic of gen-AI tools came up during the podcast, Wolpaw offered some information about his recent research into what’s possible.

“One thing we’ve been doing, and I don’t mean Valve, I mean a small group of people at Valve, we’ve been looking at some AI stuff, and I think that it’s taught me a bunch of stuff,” Wolpaw remarks. He says he doesn’t see story games succeeding in the hands of gen-AI scripts. “Like, I’m not currently worried about AI taking over creative writing because it is pretty bad at it, and I’m not just saying that defensively, like we’ve really been messing around with it.” You can watch his thoughts below, starting at the 1:09:30 mark.

Wolpaw does, however, think that there is some value in its use when it comes to simulating characters in games reacting in real time to what the player is doing. “Left 4 Dead’s a good example of, if this happens and this happens, we’ll play this line. It’s the one place where I feel like AI is worth investigating, is game writing.” He stresses that it’s not a wider Valve initiative, though acknowledges that “it’s Valve-endorsed in the sense that we are working for Valve, but we are just kind of poking around at this stuff, and this is not attached to any particular game.

“This is just some people sitting around being like, ‘This is a crazy technology. It would be kind of silly for us not to look into it at least.’ The interesting thing is that it’s not good at being especially creative, it’s not good at being funny. But imagine Grand Theft Auto where you’re going around creating a lot of physical chaos. There’s a certain amount of social chaos where you have the AI play the straight man as much as it can. And it’s just reacting to whatever insanity [is happening].”

Wolpaw says the “one thing it’s very good at is just going along with whatever insane thing you say and kind of adjusting to the flow of that. There’s something there.” He notes that developers have spent decades finding ways to make characters in games react to you, and the actions you’re performing, as naturally as possible. “This is all very sketchy, but there’s something there specifically for character interactions, and I don’t know what it is and it’s too expensive now to ship at scale, I think.”

“I want to be clear, I’m interested in this so far as it makes a better experience for people playing the game, not as a cost-cutting measure,” Wolpaw adds. “I genuinely would like to find the thing that is like, ‘This is something that could not have been done before, and opens up a whole new style of gameplay or a whole new experience.’ As just something that makes the game cheaper to make, I’m fully against it.” He also mentions testing AI voices: “Again, it’s all amazing that any of this works to me at all, but they’re so much worse than real actors.”

In summary, Wolpaw says, “I’m fascinated by it. I understand the pitfalls.” He compares it to a blend of the advent of photography and the atom bomb. “It’s something that’s both creatively destructive but could actually be literally existentially destructive.” He adds, “We’re in an industry where people are very focused on this and very against it, but when I just talk to normal people out in the world, they can’t get enough of it, which maybe is scary, too.”

Wolpaw notes he’s “maybe ten years away from the classical retirement age, but [he’s] not 20 any more. I’ve done way more than I ever thought I would. So it’s not like the industry sucks so bad that I want to be out of it, it’s just that making a game is so hard and takes so much work. If I can poke my head in and help out a little bit, I’m happy to, but I don’t know that I can devote that sort of time to it any more.”