GDC 2025 Session Viewer
Game AI Summit: The History of Game AI: From Rule-Based to Deep Learning
Youichiro Miyake (Lead AI Researcher, SQUARE ENIX)
Pass Type: All Access Pass, Summits Pass - Get your pass now!
Track: Programming
Format: Lecture
Vault Recording: TBD
Audience Level: All
There has been no introduction to AI in digital games in the 80s and 90s. I have interviewed a number of game developers and have revealed the history of AI in digital games in the 80's and 90's. I will present this information in this session. It is surely information that will be useful in future game AI development. The history of game AI begins with Pac-Man (Namco, 1980). Pac-Man has a lot to offer to the modern game AI such as meta-AI to generate slow and fast rhythms, multi-agent technology, and so on. Second, in RPG game Dragon Quest IV (Enix, 1990), the player's companion characters grow automatically. In Astronoka (Enix, 1999) is the best example of utilizing genetic algorithms and neutral net learning. From a impartial overview, the overall history of game AI is explained by showing many examples.
Takeaway
This session provides a previously unknown history of game AI. These technologies are not merely historical, but are the prototypes of modern game AI technologies. You will gain knowledge of examples that will be useful in your future development. You will learn why games excelled in the 80s and 90s.
Intended Audience
I want to share my knowledge to engineers and game designers who want to know AI technologies and want to apply them to their games. I want to take the exact history of game AI from 1980 to now to AI experts. The history shows new possibility for digital games.