what sign up
nixm nixm

← learn

What is machine learning? Software that learns from examples instead of rules

October 14, 2026

Machine learning is a way of building software where, instead of writing the rules, you show the system many examples and let it work the rules out. Show it a million emails labelled spam or not, and it learns what spam looks like. Show it years of sales, and it learns to forecast. It has quietly run your spam filter, your map, your bank's fraud checks and your video recommendations for a decade. The AI chatbots are machine learning too, just on a vastly larger scale.

The definition, and why it is a big deal

Traditional programming: a person writes rules, the computer applies them. If the rule is "flag any email with 'free money' in it", the program flags those and nothing else. Machine learning flips it: a person supplies examples and the right answers, and the computer finds the rules itself, usually rules no person could have written down. That matters because most interesting tasks have no writable rules. Nobody can list the rules for recognising a face, predicting which customers will leave, or deciding what the next word of a sentence should be. Examples exist for all of them.

How a system learns

A model starts as a shape with adjustable knobs: a formula, a decision tree, or a neural network with millions of numbers inside. Training shows it an example, lets it guess, compares the guess to the right answer, and turns the knobs slightly toward being less wrong. Repeat millions of times. When new, unseen examples also come out right, the model has learned something general rather than memorising the training set. That gap, between doing well on what it saw and doing well on what it did not, is the whole art of the field.

The three ways to learn

  • Supervised learning: examples come with the right answer. Spam or not, cat or dog, this price given these features. Most machine learning in business is this.
  • Unsupervised learning: no answers, just data, and the job is to find structure: customers who behave alike, transactions that look odd, topics in a pile of documents.
  • Reinforcement learning: the system acts, gets a reward or a penalty, and learns to act for more reward. Game-playing systems and robot control use this, and so does the final tuning step of modern chatbots, where human ratings act as the reward.

Where you already meet it

  • Your spam filter and your phone's keyboard suggestions.
  • The map's arrival time and its reroute around traffic.
  • The bank's call about a card transaction that did not look like you.
  • What the streaming app plays next, and what the shop shows you first.
  • The photo app that groups pictures by face, and the camera that brightens a dark shot.
  • Speech recognition in every voice assistant.

None of that is new. The reason "AI" became a daily word in 2022 is that machine learning at a much larger scale, applied to language, produced systems that can write and talk. The AI vs machine learning page places the terms; the neural network page explains the kind of model that scaled.

What it needs, and what that costs you

Machine learning runs on examples, and the examples come from somewhere: from people. Your emails trained spam filters; your clicks trained recommendations; and for the chatbots, a large share of what people wrote online was the training set, often along with what users type into the chat itself. That is why "does it train on my data" is the practical question to ask of any AI product. The honest answers vary. nixm's is no: nothing you type is used to train anything, and no transcript is kept.

Machine learning vs AI in one sentence

AI is the goal of machines doing things that need intelligence; machine learning is the main way we achieve it today; deep learning is machine learning with large neural networks; and generative AI is deep learning used to make new text, images and video.

frequently asked

What is machine learning in simple terms?

Software that learns a task from examples instead of from rules a programmer wrote. Show it enough labelled examples and it works out the pattern, then applies it to new cases.

What is the difference between AI and machine learning?

AI is the broad goal of machines doing intelligent things. Machine learning is the main current method for getting there, by learning from data. All machine learning is AI; not all AI is machine learning.

What are the types of machine learning?

Supervised, where examples come with answers; unsupervised, where the system finds structure in unlabelled data; and reinforcement, where it learns from rewards for its actions.

Where is machine learning used in everyday life?

Spam filters, keyboard suggestions, map routing, fraud detection, streaming and shopping recommendations, photo organisation, and speech recognition.

Is ChatGPT machine learning?

Yes. It is a very large neural network trained on text, which is deep learning, a kind of machine learning, with a final tuning step that uses human ratings as a reward signal.

Does machine learning use my data?

Often, yes: products learn from what their users do and type. Check whether an AI tool trains on your input. nixm does not, and keeps no transcript.

try it now — no account, no transcript kept. even the signal dissolves in seven days.

seed a signal →

Enter your email and receive a link to sign in. No passwords, just a secure signal to connect.

why connect

  • 20 free credits to start
  • your signals, remembered
  • buy more credits anytime
  • image & video generation features
or
+