Machine Learning
Chapter 3 · page 58
Definition
Machine learning is a branch of artificial intelligence that lets a system learn from data rather than being programmed rule by rule. We give it examples, and it automatically adjusts its parameters to fit them. From them it extracts statistical patterns that it can reuse on new data. It is currently the dominant approach to designing high-performing AI systems.
Example
To train a speech recognition system, we have it listen to thousands of hours of recordings along with their transcripts. The model learns to match sounds to words, and can then transcribe sentences it has never heard, because it has picked up the statistical patterns of language. This type of learning is generally supervised: each example comes with the “right answer”, which guides the model as it adjusts.
Why it matters
Nearly all of today’s AI relies on machine learning: facial recognition, spam filters, recommendations, voice assistants, and so on. This is a profound shift: we no longer describe the solution explicitly, we let the model discover it from the data. These systems, however, lack human-like understanding; they produce results from calculations and probabilities. That distinction helps us to grasp both their power… and their limits.
How to cite this text
Vial, S. (2026). Machine Learning. In A Living Lexicon of Artificial Intelligence (p. 58). Stéphane Vial, publisher, Montréal.
