Full answer
Traditional software follows explicit instructions: if X happens, do Y. Machine learning flips this. Instead of writing the rules, you feed a system thousands or millions of examples and let it discover the patterns on its own. A spam filter trained on machine learning doesn't have a list of banned words; it infers what makes an email spammy by analyzing millions of labeled examples.
There are three main learning styles. Supervised learning trains on labeled data — showing the model 10,000 photos labeled "cat" or "not cat" until it can classify new photos on its own. Unsupervised learning finds hidden structure in unlabeled data, like grouping customers into segments without being told what the groups should be. Reinforcement learning trains an agent through trial and error, rewarding good actions and penalizing bad ones — this is how game-playing AIs like AlphaGo were built.
Machine learning powers most of the technology people use daily: the recommendation algorithms on Netflix and Spotify, the voice recognition in Siri and Alexa, fraud detection in banking apps, autocorrect on your keyboard, and the image recognition that lets your phone unlock with your face.
The term is often used interchangeably with "AI" in the press, but they're not the same thing. AI is the broader field encompassing any attempt to make machines intelligent. Machine learning is one specific approach within AI — currently the dominant and most successful one. Deep learning is a subset of machine learning that uses neural networks with many layers, and it's the engine behind modern image recognition and large language models.
You don't need to be a mathematician to use machine learning tools today. Platforms like Google AutoML, Hugging Face, and Apple's Create ML let non-experts train and deploy models with minimal code.