You don't need a PhD to understand what machine learning is and how it works. You do need a good mental model.
The Core Idea
Traditional programming: you write rules → computer follows them.
Machine learning: you give the computer examples → it figures out the rules.
That's it. Everything else is implementation detail.
Supervised Learning
The most common type. You show the model thousands of labelled examples ("this is a cat", "this is not a cat") and it learns to recognise patterns.
Real-world examples: spam filters, credit scoring, image recognition, recommendation engines.
Unsupervised Learning
No labels. The model groups data based on similarity. Useful when you don't know what you're looking for.
Real-world examples: customer segmentation, anomaly detection, topic modelling.
Neural Networks in One Paragraph
A neural network is loosely inspired by the brain. It has layers of nodes. Data passes through each layer, getting transformed slightly. The final layer outputs a prediction. During training, the network adjusts its internal values to reduce its error rate. After enough examples, it gets good.
Why Now?
Three things converged: massive datasets (the internet), cheap compute (GPUs, cloud), and better algorithms (transformers). The result is tools like GPT, Stable Diffusion, and AlphaFold — things that seemed like science fiction five years ago.
Where to Start
- Play with Google's Teachable Machine — no code, instant results
- Take a Python ML course to understand the code layer
- Build a small project: train a classifier on a Kaggle dataset
The math comes later and makes more sense once you've seen what it's describing.