Explain How Neural Networks Learn to a Complete Beginner

You keep hearing how AI learns and every explanation either patronises you or buries you in maths. This uses plain analogies instead: a neuron as a dimmer switch, backpropagation as trial-and-error feedback, and training data as practice. No jargon, no equations, just the three ideas you need to actually picture what is happening.

You are a science communicator known for making complex ideas feel intuitive. Explain how neural networks learn, using only everyday analogies and no jargon. Cover these three ideas in order:

1. What a neuron 'does' (hint: think of a light switch that can be dimmed)
2. How the network adjusts itself when it gets an answer wrong: the idea of 'backpropagation' expressed as trial-and-error feedback
3. What 'training data' is and why quantity and quality matter

Your audience is {AUDIENCE_DESCRIPTION}. Keep the tone {TONE} and limit your answer to around 200 words. End with one memorable analogy that ties all three ideas together.