Take a Load Off! Enhancing Artificial Intelligence Offloading in Multiplayer Games


  • Prajakta Gadhamsetty Aspiring Scientists' Summer Internship Program Intern
  • Sai Vaka Aspiring Scientists' Summer Internship Program Intern
  • Stephanie Kane Aspiring Scientists' Summer Internship Program Co-mentor
  • James Casey Aspiring Scientists' Summer Internship Program Mentor




In online multiplayer games, the architecture typically consists of a central game server and clients, which display the game while sending and receiving data to and from the game server. The majority of processing is typically done on the client side to lighten the load on the server. We can improve these games by creating more complex AI by using AI offloading for greater computational efficiency. The idea is that the AIs’ logic is computed by different machines besides the game server. Offloading AI allows developers and even players to create smarter, more complex AI, as the game isn’t bottlenecked by one main server. These AI can easily be added or removed from the game, as they are treated similarly to a player. However, this method has drawbacks, including security threats, latency caused by the communication between the AI machine and servers, and AI machines crashing or disconnecting. These machines would need to be treated differently than a player, with checks that automatically restart the machine or have fallback AI in case of a malfunction. In this paper, we will be exploring how to mitigate these drawbacks and find use cases for the technique.





College of Visual and Performing Arts