EA's SEED lab built AI agents that can play through EA games on their own to find bugs, exploits, and broken navigation paths — acting like tireless automated playtesters. The system has been used to test Battlefield and Dead Space games, where manual testing would take hundreds of thousands of hours.

The agents use reinforcement learning (a training method where an AI learns by trial and error, earning rewards for useful actions) and imitation learning (where an AI learns by studying recordings of human players). Specific methods include CCPT (Curiosity-Conditioned Proximal Trajectories) for bug-finding and MultiGAIL (Multimodal Adversarial Imitation Learning) for simulating different player styles. In Battlefield 2042, an AI vehicle-driving agent using path-following algorithms achieved a 39% reduction in navigation time compared to earlier scripted bots. Battlefield V alone requires testing 601 distinct features, estimated at 0.5 million hours of work annually.