Engaging Bots for Multiplayer Games
summary
Modl.ai maps real player replays from Forge and Fight! into an AI engine that trains bots to craft player-like weapons and fight with human timing. The game runs in Unity on Photon Quantum, letting the team extract deterministic, frame-by-frame replays for training. A hybrid stack uses a finite-state top-level and imitation learning for low-level combat, needs <10 hours of gameplay data to train, and yields a 50% increase in day-one retention.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 1:37Overview
- 2:45What is Fortune Fight
- 4:31Fortune Fight Gameplay
- 5:45What the devs were looking for
- 7:15The right game feel
- 7:30The tech stack
- 11:16The paradigm
- 12:55imitation learning
- 13:34data collection
- 13:57hybrid approach
- 15:39side by side
- 16:11Fortune Fight
- 16:41Takeaways
- 17:19What else could you do
- 17:49Long term
- 18:16Recap
- 18:36Outro
related talks
48mMachine Takeover - Improving Your Game With Machine Learning & AI
37mA practical perspective on training reinforcement learning bots.
01h 01mIntroduction to Practical AI
32mThe Road to General AI
21mBalancing Player Experience Without Players At Supercell: Case mo.co
52mPanel: Playing Together How To Boost Retention by Adding World-class Multiplayer Options
52mThe Journey of AI in Games: A Technical and Philosophical overview
20mStudio Speaks: The Reality of AI Adoption in Games