Clicked Gallery

Machine learning vs AI: what's the difference?

Highlighted from a real engineering doc. Explained by Clicked.

Used in a sentence

The team replaced its hand-written rules with a machine learning model trained on two years of labelled email.

The reader highlighted one word in the docs. Clicked explained the technical term “machine learning” in simple terms:

Explained in three depths

●○○

Overview

▾

Artificial intelligence is the goal: getting machines to do things that normally require human judgement. Machine learning is the method that now dominates, where a program works the rules out from examples instead of being given them. All machine learning is AI. Not all AI is machine learning.

●●○

Detail

▾

The two words sit at different levels: one names an aim, the other a way of reaching it. Artificial intelligence is the aim, and it begins where a system is built to make judgement calls a person would otherwise make. For decades that meant experts writing their judgement down as rules, which is how early medical diagnosis systems worked. Machine learning reaches the same aim from the other direction: show a program a hundred thousand emails already marked spam or not, and let it find the pattern itself. Why did that approach take over? Because for most interesting problems the rules are endless, shifting and impossible to state, which holds for spam, translation and medical scans alike. The catch is that a system which learned from examples cannot tell you what it learned, and it absorbs whatever bias sat in them. Neither approach covers your calculator, which follows instructions exactly and exercises no judgement at all. AI is the aim, machine learning is the branch that reaches it through data, and that is why the words are not interchangeable.

●●●

Analogy

▾

Transport and cars. Transport is the goal, which is getting from one place to another. The car is the method that came to dominate, though bicycles, trains and boats are transport too. Calling every vehicle a car would be wrong in exactly the way that calling every AI system machine learning is wrong.

AI explanations may contain errors · Not professional advice

Same facts, different vibe — Slang mode 😎

●○○

Overview

▾

AI is the ambition: get a machine to make the kind of call a person would normally have to make. Machine learning is the technique that took over, where the thing works out the rules from examples rather than being handed them. Every bit of machine learning is AI. Plenty of AI never involved learning at all. 😎

●●○

Detail

▾

AI means getting a machine to make a call that a person would otherwise have to make. Machine learning is one route there: show a program enough labelled pictures and it works out the difference itself, which is why your photos app finds every shot of your dog. Nobody wrote down what a dog looks like in code, because that description does not usefully exist. The older route was people writing expert rules out by hand, and that counted as AI too. Why did learning from data win? Because the rules for most real problems are endless and keep moving, and nobody was finishing that list. Price of admission: your photos app cannot explain what a dog looks like to it, and label enough of those pictures goat by mistake and it will cheerfully report that you own eleven goats. Your thermostat, meanwhile, makes no judgement whatsoever and is not AI, whatever the box says. 😎

●●●

Analogy

▾

Curing illness is the goal. Antibiotics were the method that changed everything, to the point where people started saying medicine and picturing pills. Surgery, physiotherapy and vaccines are all still medicine. Machine learning is the antibiotic here: the arrival that worked so well it quietly borrowed the name of the whole field. 😎

AI explanations may contain errors · Not professional advice

Formal definition — The same term, explained the usual way

Artificial intelligence is the field concerned with systems that perform tasks ordinarily requiring human judgement, encompassing symbolic and rule-based approaches as well as statistical ones. Machine learning is the subfield in which behaviour is derived from data through a training process rather than specified explicitly. It presently accounts for most deployed AI systems, though deterministic software that merely executes specified instructions falls outside the field entirely.

Want Clicked to explain terms like “machine learning” directly in your browser — including on PDFs?

Add to Chrome — Free

50 free Explanations · No credit card required

More from the gallery

What is a Neural Network?

Math that learns the rules from examples instead of having them written, and why that wins.

What is a large language model (LLM)?

A program that predicts what text comes next, then taught separately to answer rather than ramble.

What is training (in AI)?

The process that builds a model by example. Chatting with it afterwards changes nothing.

Overfitting vs. Underfitting: What's the Difference?

Memorized the answers vs. never opened the book — and the score gap that exposes both.

What is the Bias-Variance Tradeoff?

Too rigid vs. too impressionable — and why total error is lowest in between.

What is Synthetic Data?

Training data made by machines, why labs increasingly rely on it, and how it goes wrong.