AI for Gaming: How I Built a Bot to Play a Video-Game with R and Python
summary
A bot that plays the browser puzzle Christmas Three achieves a mean score of 42.5 per move (creator scores ~20), implemented with reinforcement learning + custom computer vision trained on reduced subboards (windowing, transposition, single-tile strategies) up to 5×4 before convergence fails. The system uses R/Python, Posit Cloud, magick and multinomial logistic regression for tile classification, and Selenium + Chrome for GUI automation and an in-page assistant; development takes about three weeks.
AI-written summary · about
chapters (14)
- 0:00AI for gaming intro
- 1:06Christmas Three mechanics
- 2:55State space in RL
- 4:22Reducing state space
- 7:41Training model results
- 8:45Game interaction via Selenium
- 10:32AI performance evaluation
- 11:58Business and project goals
- 13:29Generative AI for customization
- 14:12Data science education
- 16:41Future directions
- 17:22Q&A: AI vs human
- 17:46Q&A: Adapting to other games
- 19:13Q&A: Project timeline
