20.3 AI in Finance

AI in finance powers fraud detection, credit scoring, algorithmic trading, and customer-service chatbots. Because financial data is structured and abundant, models can learn subtle patterns — for example, the signature of a fraudulent transaction hidden among millions of normal ones.

Anomaly detection: unusual transactions stand out from normal clusters.

When you pay with a mobile-money wallet like telebirr and a transaction suddenly comes from a new device in another city, an anomaly-detection model can flag it in real time and ask you to confirm before the money moves.

Scenario

Your fraud model has 99.9% accuracy but misses most actual fraud. What should you do?

Check your understanding

1/4 · 40 XP

Fraud detection is usually framed as which kind of problem?