Probability, AI and the Modern World
Probability plays a central role in how modern systems operate, from artificial intelligence to scientific research and public policy. Rather than offering certainty, these systems rely on likelihoods and patterns to make predictions.
- Probability supports modern decision-making
- Used in technology, science and social systems
- Highlights the limits of certainty in complex environments
Data to Decisions
Modern systems increasingly rely on data-driven predictions.
However, predictions are not the same as decisions.
Probability estimates what is likely, but humans must still understand how to act on that information.
- Prediction expresses likelihood, not truth
- Decisions involve values and judgement
- Confusing prediction with certainty can be dangerous
Conditional Probability
Conditional probability describes how the likelihood of an event changes when new information is known.
Context matters, and probabilities are rarely fixed.
- Given that "X" has happened, how likely is "Y"?
- New information reshapes expectations
- Everyday reasoning often uses conditional probability without us realising
Bayes' Theorem
Bayesian thinking is a formal way of updating beliefs when new evidence appears. Instead of starting from scratch, we adjust what we already believe based on how strong the new evidence is.
- Start with an initial belief (prior)
- Add new evidence
- Update the belief gradually
- Strong evidence matters more than weak evidence
Example of Bayes' Theorem: Monty Hall
The setup
Imagine you are on a game show. There are three doors. Behind one door is a fantastic prize (let's say a new bike!), and behind the other two are socks.
- You pick a door: you choose one door, say Door 1. At this point, you have a 1/3 chance of being right and winning the bike.
- The host knows: the host knows where the bike is. They then open one of the other doors that you didn't pick, revealing socks. For example, if you picked Door 1, the host might open Door 3 and show you socks.
The choice: to switch or not?
The host then asks, "Would you like to stick with your first choice (Door 1) or switch to the remaining closed door (Door 2)?"
What do you think you should do?
Most people think it doesn't matter, and that it's a 50/50 chance now that there are only two doors left. But that's not true!
The explanation
You should always switch! Here's why:
- Your first choice: when you first picked a door, you had a 1/3 chance of being right.
- The other doors: that means the other two doors combined had a 2/3 chance of hiding the bike.
When the host, who knows where the prize is, eliminates a "socks" door, they are concentrating that full 2/3 probability onto the single remaining closed door. The 1/3 probability of your original choice never changes.
So, by switching, you move from a 1/3 chance of winning to a 2/3 chance of winning!
Why Does Bayesian Thinking Even Matter?
Bayesian thinking helps prevent overreaction to new information.
It explains why surprising results should not always change beliefs dramatically, especially when the events are quite rare.
- Base rates matter
- Rare events are often misunderstood
- Encourages rational belief
- Helps explain medical test errors
Probability in AI
Artificial intelligence systems use probability to recognise patterns and make predictions from big datasets.
AI does not understand meaning as we humans do. It calculates which outcomes are most likely to happen.
- AI estimates likelihoods
- Outputs are probabilistic, not truth
- Predictions depend heavily on training data
Confidence vs Correctness in AI
High confidence from an AI system does not guarantee correctness.
Models can be wrong, especially when data is incomplete or biased.
- Confidence is not the same as accuracy
- Bias in data leads to biased predictions
- AI reflects the uncertainty in the real world
Ethics and Misuse of Probability
When probabilistic predictions are treated as facts, ethical problems arise.
This is especially serious in areas like criminal justice, hiring and healthcare.
- Predictions can reinforce bias
- Statistics can be used to justify unfair decisions
- Human judgement must remain involved
How This Topic Links to Game Theory and the Butterfly Effect
Statistics and probability provide the foundation for understanding uncertainty in complex systems, which directly connects to both Game Theory and Chaos Theory. In Game Theory, players make strategic decisions based on probabilities, expectations and incomplete information about others' actions. Outcomes emerge from many individual choices, often producing results that cannot be predicted with certainty.
Similarly, the Butterfly Effect in Chaos Theory shows how little changes in initial conditions can compound into large, unpredictable outcomes. Probability helps explain why precise long-term prediction is impossible in such systems, even when the underlying rules are known. Together, these ideas show that in economic, social and natural systems, rational decision-making must account for uncertainty, feedback loops, and the possibility that small actions can have disproportionate effects. Chaos Theory also emphasises how beliefs have to be updated and changed.
Part 3 Takeaway
Probability is not just a mathematical tool, but a way of thinking about evidence, uncertainty and decision-making. It shapes how we understand truth, technology and responsibility in a complex world.
- Encourages careful reasoning
- Supports ethical decision-making
- Connects mathematics to society and philosophy