Brian Raffel’s Post

This is a timely question! I recently shared my excitement about AI, particularly reinforcement learning, in video games. As an RTS/RTT fan, I see huge potential for smarter AI opponents. For example, in Total War games like Warhammer III's Immortal Empires, the mix of strategic campaigns and tactical battles often suffers from exploitable or "cheating" AI. RL-based AI could add depth and unpredictability to these experiences. As a longtime Call of Duty fan (Call of Duty 3), I’ve been playing the Black Ops 6 Zombies mode extensively lately and I imagine smarter, RL-driven enemies could add so much depth and fun to the game. Instead of just more and tougher zombies, instead there are zombie types that adapt—learning optimal map navigation, targeting key players (those that revive the most), or striking when least expected. The agent powering the zombies would be trained on the vast data collected by the millions of matches played by players.

Offline Multiplayer Large Game Modes Using AI to create a Learning Model that allows for Multiplayer large modes to feel authentic and alive. Allows gaming to expand beyond readily available internet