Too Clever isn't Fun
summary
Rival Stars uses a simplified 5v5 model with eight selectable plays and player-versus-match performance modifiers that can shift success chance by up to 30%. The AI brute‑forces every known play, simulates outcomes, scores expected yardage, then picks via weighted random selection so choices feel sensible but sometimes surprising. To avoid “cheating” it only inspects plays already seen, limits reaction by using a perceived velocity vector for shot leading, and fudges grenade predictions so pressure feels cinematic without breaking believability.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 4:06Human-like AI in games
- 7:22AI basics for football
- 11:58AI design in Rival Stars
- 15:27Balancing AI difficulty
- 19:51Making AI feel fair
- 23:45Boss fight design
- 33:07Machine learning vs human-like AI
- 41:12Cheating subtly in boss health bars
- 49:05Q&A
related talks
59mGame AI 4N00bz
28mMake It Fast: Simultaneous AI Actions in Gears Tactics
01h 01mIntroduction to Practical AI
27mCan computers become intelligent through playing games?
46mBeating Go thanks to the power of randomness
01h 18mTotal War: how AI can deliver entertainment.- Duygu Cakmak & Guy Davidson [ACCU 2018]
39mEmbarrassing Mistakes: a short story about AI
34mAn AI assistant for football tactics