Conditional Probability

$P(A|B)$ means, probability of $A$ given $B$.

In ML contexts, models usually calculated $P(Y|X)$ where $X$ is input features and $Y$ is the model prediction.

For example in a spam detector model predicts $P(\text{spam} | \text{email})$ in other words, "probability of email being a spam email, given there's an email"