Clinical AIACADEMY
The learning path

MODULE 2 / 8

How AI Learns (and Why It Sometimes Gets It Wrong)

Jump to

Section 1 of 9

In this module

Topics to explore:

  • Explain how AI models learn patterns through training on labelled examples
  • Define overfitting and understand why it undermines generalisation
  • Identify warning signs that a model may have learned non-clinical shortcuts
  • Understand the importance of diverse, high-quality data
  • Evaluate whether an AI model is likely to work in new patient populations

Slide 1 – Let’s Recap

From Module 1, you learned:

  • AI learns from examples, not rules
  • It builds patterns—not understanding
  • It performs well only if the training data was fair and relevant

Now, we’ll explore how it learns, and what can go wrong inside that black box.