The computer must constantly choose between using moves it already knows work and trying completely new ones.

This is called the exploration-exploitation trade-off. Think of it like deciding where to eat dinner. Do you go to your favorite restaurant because you know the food is great? Or do you try a brand-new place that might be even better, but could also be terrible? The AI faces this exact dilemma at every step. It must balance sticking to safe, known rewards with exploring new paths to find even bigger payoffs.