Introduction to Practical AI
summary
Clear, concrete examples of game AI and what actually worked: store recommendations use clustering to avoid offering a Teemo skin to players who hate Teemo; QA automation cites Battlefield V needing 601 features and ~500,000 hours (≈300 work years) of manual testing, so RL plus imitation learning expands coverage but requires millions of runs; GAN up-res mods revive Mass Effect textures; a home project commercial quieter reaches ~92% accuracy; an auto-skinning hack fails from 400 training spheres and overfitting. Legal notes: Adobe Firefly opts-in training, GitHub Copilot faces litigation.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 1:45AI and ML basics
- 5:17How ML learns patterns
- 9:51Machine learning in video game development
- 13:43ML challenges and limits
- 15:06Neural networks explained
- 16:08AI for game testing
- 21:25Generative adversarial networks explained
- 27:09Generative AI applications
- 32:02AI project tips
- 42:52AI ethics and legal issues
- 49:55Conclusion
related talks
52mThe Journey of AI in Games: A Technical and Philosophical overview
29mThe Present & Future of AI and Games
32mThe Road to General AI
39mThe Pocket Guide to Artificial Intelligence
48mMachine Takeover - Improving Your Game With Machine Learning & AI
02mDesigning a Video Game AI
19mNow in AI - How We Got Here
41mThe Current State Of AI In Games