Valve's Deadlock is currently listed as being in 'early development'; however, it's one of the most popular games on Steam. Deadlock blends hero-shooter elements from games like Team Fortress 2 and Overwatch with the lane-based action of a MOBA like DOTA 2. It's a mix of styles and mechanics that feels familiar and fresh, so it's not out-of-step to say that Deadlock is shaping up to be another hit from a studio that releases very few games.
Valve is no stranger to multiple games or shooters; Deadlock is a third-person shooter with several decades of running servers and handling matchmaking for some of the most beloved PC games ever made. So you might think this would be one side of Deadlock that it wouldn't need any outside help to sort out. Especially not ChatGPT.
ChatGPT is becoming an invaluable tool for many, and Valve engineer Fletcher Dunn used the generative AI chatbot to help out with Deadlock's matchmaking system.
As detailed in his post on X, Fletcher described a hypothetical matchmaking algorithm that would solve what he and the team were looking for. ChatGPT not only understood the hypothetical scenario but also provided a solution - something called the Hungarian algorithm. And with that, Deadlock's matchmaking and hero selection to the Hungarian algorithm.
This is an excellent example of a game developer using generative AI to find a solution to complex problems quickly and cost-effectively. Fletcher isn't alone either; it's estimated that one in three gamer developers, programmers, and engineers uses generative AI. Whether that's as a tool to find the most suitable algorithm for multiplayer matchmaking or creating a digital human that players can interact with, AI is already significantly impacting how games are made.