Algorithm

A set of rules or instructions given to an AI, neural network, or other machines to help them learn on their own.

How it works

An algorithm is a finite, ordered sequence of well-defined steps that transforms an input into an output. In machine learning, training algorithms iteratively adjust model parameters to minimise a loss function — for example, gradient descent computes the gradient of the error and nudges weights in the direction that reduces it. The algorithm terminates when a stopping criterion is met, such as the loss falling below a threshold or a maximum number of iterations being reached.

Why it matters

Every AI capability — from image recognition to language generation — is ultimately the product of an algorithm running on data. Breakthroughs in algorithm design (backpropagation, attention, RLHF) have historically produced larger performance leaps than hardware improvements alone. Understanding algorithms gives practitioners the ability to diagnose failures, optimise performance, and build genuinely novel AI systems rather than treating them as black boxes.