A bag holds 5 marbles: 2 red and 3 blue. Draw one and look: it is red. Keep it out. Now the bag holds 4 marbles and only 1 is red, so the chance the next marble is red has dropped from 52 to 41. The first draw changed what is left, and that is dependence: the second event's probability depends on what happened first.
Had the first marble been blue, the bag would hold 2 red out of 4, and the chance of red would rise to 42=21. The second probability moves with the first result, which is why it is written P(B∣A), read 'the probability of B given A'. To find the chance that both happen, multiply the first probability by the updated second one: P(red, red)=52×41=101.
Put the marble back before the second draw and the bag is restored: the second draw is red with probability 52 whatever came first. That is independence, and only then is P(A)×P(B) correct. Dependence is also weaker than causation. Umbrella sales and rained-out baseball games are dependent, yet umbrellas do not cancel games; rain drives both.
Think of it like this. Dependent events are like sharing a pizza. Once someone takes a pepperoni slice, fewer pepperoni slices remain, so the next person's chance of getting one drops. If slices were put back (not recommended), the next person would face the original pizza, which is independence. The pizza fits draws without replacement best; dependence can also come from a shared cause, such as weather.
Core idea. Two events are dependent when P(B∣A)=P(B) — learning A happened shifts B's probability.