A Valve developer recently leveraged ChatGPT to significantly improve Deadlock's matchmaking system. This follows considerable player criticism of the previous MMR-based system, with complaints focusing on uneven team skill levels and frustrating matches. According to engineer Fletcher Dunn, ChatGPT suggested the Hungarian algorithm as a solution, a recommendation he implemented.
Dunn's Twitter posts detail his interaction with ChatGPT and highlight the AI's effectiveness. He's even dedicated a Chrome tab solely to ChatGPT, emphasizing its increasing utility in his workflow. However, he also acknowledges a potential downside: the reduced need for human interaction, either in person or via online discussions. This sparked debate, with some commentators expressing skepticism about AI replacing human programmers.
The Hungarian algorithm, a type of bipartite matching algorithm, addresses matchmaking by prioritizing one side's preferences (in this case, player preferences). This contrasts with the previous system, which apparently failed to adequately balance team skill. The new algorithm aims to match players with more suitably skilled teammates and opponents.
Despite the improvement, some Deadlock players remain unhappy, voicing continued frustration with the matchmaking experience. Their negative reactions underscore the ongoing challenge of achieving perfect matchmaking, even with advanced algorithmic solutions. Nevertheless, the Deadlock team's use of ChatGPT demonstrates the evolving role of AI in game development and highlights its potential to accelerate problem-solving.
