IB AI: Probability and Expected Outcomes

Probability is the language of uncertainty, and expected value is its interpreter. In IB AI, you’ll move beyond simple chance to model real-world systems—from game design to financial risk—by quantifying likelihoods and predicting long-term outcomes. Mastering these concepts gives you the analytical tools to make informed decisions in the face of randomness.

Foundational Probability: Sample Space and Probability Types

Every probability problem begins by defining the sample space, which is the set of all possible outcomes of a random experiment. For a single six-sided die, the sample space is $S=\{1,2,3,4,5,6\}$. The probability of an event is a number between 0 and 1 that quantifies how likely it is to occur.

You will work with two primary types of probability. Theoretical probability is determined by reasoning about the experiment's structure. If all outcomes in a finite sample space are equally likely, the probability of event $A$ is: $$P(A)=\frac{\text{Number of outcomes in }A}{\text{Total number of outcomes in the sample space}}$$ For example, $P(\text{rolling an even number})=\frac{3}{6}=0.5$.

In contrast, experimental probability (or empirical probability) is determined by repeating an experiment many times and observing the relative frequency. If you flip a coin 1000 times and get 523 heads, the experimental probability of heads is $\frac{523}{1000}=0.523$. As the number of trials increases, experimental probability typically converges toward the theoretical probability, a principle known as the Law of Large Numbers.

Modeling Combined Events

Many situations involve more than one event. We describe the relationship between events $A$ and $B$ using specific terminology and visual tools. The union $A\cup B$ means "A or B occurs." The intersection $A\cap B$ means "both A and B occur." Events are mutually exclusive if they cannot occur at the same time (e.g., rolling a 1 and rolling a 6 on a single die).

A tree diagram is an indispensable tool for visualizing multi-stage experiments. Each branch represents a possible outcome at a stage, and its associated probability is written on the branch. To find the probability of a sequence of events, you multiply the probabilities along the branches that lead to that outcome. The probabilities of all final outcomes sum to 1. For instance, to model flipping a fair coin twice, your first stage has branches for Heads (0.5) and Tails (0.5). Each of these branches then splits again into a second Heads (0.5) and Tails (0.5) branch, leading to four outcomes each with probability $0.5\times 0.5=0.25$.

Understanding Conditional Probability

Conditional probability is the probability of an event occurring given that another event has already occurred. It is denoted $P(A|B)$, read as "the probability of A given B." The formula is: $$P(A|B)=\frac{P(A\cap B)}{P(B)},\text{provided }P(B)>0$$

Imagine a deck of cards. The probability of drawing an Ace (event A) is $\frac{4}{52}$. However, if you know the card is a Heart (event B), the probability changes. Here, $P(A\cap B)$ is the probability of the Ace of Hearts: $\frac{1}{52}$. $P(B)$ is the probability of a Heart: $\frac{13}{52}$. Thus, the conditional probability is $P(A|B)=\frac{1/52}{13/52}=\frac{1}{13}$. This concept is crucial for revising predictions based on new information, a process central to Bayesian statistics and machine learning algorithms.

Calculating and Applying Expected Value

When outcomes have associated numerical values (like a score or a monetary payoff), we move from probability to expected value. For a discrete random variable $X$, which takes values $x_1,x_2,...,x_n$ with corresponding probabilities $p_1,p_2,...,p_n$, the expected value $E(X)$ is the long-run average value if the experiment is repeated infinitely many times. It is calculated as: $$E(X)=\sum _{i=1}^n{x}_i\cdot p_i=x_1{p}_1+x_2{p}_2+...+x_n{p}_n$$

Consider a simple game: you pay \$2 to roll a die. If you roll a 6, you win \$10; otherwise, you win nothing. Your net gain, $X$, can be -\$2 (if you lose) or +\$8 (if you win, which is \$10 minus the \$2 cost). $P(X=8)=\frac{1}{6}$ and $P(X=-2)=\frac{5}{6}$. The expected net gain is: $$E(X)=8\times \frac{1}{6}+(-2)\times \frac{5}{6}=\frac{8}{6}-\frac{10}{6}=-\frac{2}{6}\approx -\$0.33$$ A negative expected value indicates a losing game for the player in the long term. This calculation is the bedrock of analyzing games of chance, setting insurance premiums (where the premium must exceed the expected payout of claims), and decision-making scenarios under uncertainty (like choosing an investment with the highest expected return, adjusted for risk).

Common Pitfalls

1. Confusing Mutually Exclusive and Independent Events: Mutually exclusive events cannot happen together ($P(A\cap B)=0$). Independent events do not affect each other's probability ($P(A|B)=P(A)$). These are very different properties. A common error is assuming that because events are mutually exclusive, they are independent. In fact, if they are mutually exclusive and one occurs, the other definitely does not, so they are dependent.

1. Misapplying the Multiplication Rule: The probability of $A$ and $B$ is $P(A\cap B)=P(A)\times P(B|A)$. A frequent mistake is using $P(A)\times P(B)$ in all cases, which is only valid if $A$ and $B$ are independent. Always check for independence before using the simplified multiplication rule.

1. Misinterpreting Expected Value: Expected value is a long-term average, not a prediction for a single trial. An expected net gain of +\$0.50 per lottery ticket does not mean you win money on this ticket; it means if you bought thousands of tickets, your average gain per ticket would be about 50 cents. It does not describe the variability or risk involved.

1. Ignoring the Sample Space in Conditional Probability: When calculating $P(A|B)$, your universe shrinks to the outcomes where $B$ is true. A classic error is to keep the original, larger sample space in the denominator. Always remember: "Given B" means B becomes your new, restricted sample space.

Summary

• Probability Fundamentals: The sample space is your starting point. Theoretical probability is derived from logic, while experimental probability is based on observed data, converging to the theoretical value over many trials.
• Modeling Complex Events: Use set notation (union, intersection) and tree diagrams to systematically break down and calculate probabilities for multi-stage or combined events.
• Conditional Reasoning: Conditional probability $P(A|B)$ quantifies how the probability of event $A$ changes with the knowledge that $B$ has occurred, calculated by $P(A\cap B)/P(B)$.
• Predicting Long-Term Averages: The expected value $E(X)$ of a discrete random variable is the sum of each outcome multiplied by its probability ($\sum x_i{p}_i$). It is the cornerstone for analyzing the fairness of games, pricing insurance, and informing rational decisions under uncertainty.