I built a Blackjack playing AI that learns entirely through experience by playing thousands of simulated games. Instead of relying on fixed rules, the agent learns when to hit or stick by observing which actions lead to better outcomes over time. As training continues, it begins to form strategies that closely match optimal Blackjack play, even in more challenging situations such as strong dealer cards. By the end of training, the agent reaches a stable win rate close to the theoretical best, showing how reinforcement learning can be used to train an AI to make effective decisions.
