The Kaplan-Meier estimator calculates the probability of surviving in a given length of time while considering the presence of censored observations. The survival probability is updated at each event (e.g., death, failure), rather than at regular intervals, making it a step function that changes value only at the time of each event.