A Valve developer recently leveraged ChatGPT to significantly improve Deadlock's matchmaking system. Facing criticism for its previous MMR system, the Deadlock team, using ChatGPT's suggestion, implemented the Hungarian algorithm. This followed player complaints about unevenly matched games, with experienced players frequently facing less skilled teammates.
The developer, Fletcher Dunn, documented his ChatGPT interaction on Twitter, showcasing the AI's role in recommending the Hungarian algorithm, specifically highlighting its suitability for scenarios where only one party (e.g., a player) expresses preferences. Dunn's tweets highlight both the power and potential downsides of using generative AI, noting that while it speeds development, it might reduce human interaction and collaboration. Some critics expressed concern that AI tools like ChatGPT could replace human programmers, a sentiment Dunn partly acknowledged.
The Hungarian algorithm, a type of bipartite matching algorithm, is designed to find optimal pairings given specific constraints. In Deadlock's context, this likely means matching players based on skill level and preferences, leading to fairer and more balanced gameplay.
Despite the improvement, some Deadlock players remain unhappy with the matchmaking, expressing frustration on social media. This highlights the ongoing challenge of satisfying a diverse player base, even with advanced algorithmic solutions. Nevertheless, the Deadlock team's innovative use of ChatGPT underscores the evolving role of AI in game development. The successful integration of the Hungarian algorithm, facilitated by ChatGPT, represents a noteworthy advancement in Deadlock's development.
