Skip to main content
Mobius
Intermediate

What is machine learning

Also known as: ML

Definition

Machine learning is the part of AI where software learns rules from examples instead of being told them, then uses those rules to predict or sort new cases.

Why it matters

Machine learning sits inside many tools a small business already uses: spam filters, card fraud alerts, "recommended for you" and quantity suggestions in till and shop software.

Common mistakes

Expecting a model to learn from a few dozen rows, when it needs enough examples of each situation.

Related terms

Worked exampleIllustration

A fictional neighbourhood bakery forecasts Friday challahs

A fictional neighbourhood bakery with sales records for 52 Fridays. The numbers are invented for this example.

  1. The old rule

    "Bake what we sold last Friday." After a holiday eve that means 360 challahs for an ordinary Friday that sells about 120, so 240 are left over.

    360 - 120 = 240

  2. The examples

    The till software gets 52 past Fridays with the date, holidays, rain and the number sold: 39 ordinary dry Fridays, 8 rainy ones and 5 holiday eves.

    39 + 8 + 5 = 52

  3. The pattern

    It finds that a holiday eve sells about three times an ordinary Friday, about 360 against 120, and a rainy Friday about 100.

    3 × 120 = 360

  4. The forecast

    For next Friday, ordinary and dry, it suggests a range, not one number.

    110 to 130

  5. The check

    The owner bakes 120 and sells 116. That Friday becomes the 53rd example for the next retraining.

    120 - 116 = 4

The rule came from a person, the pattern came from the examples, and both need checking.

Illustration

52 Fridays teach the forecast

Learned from 52 Fridays, the forecast says to bake about 120 on an ordinary Friday, about 100 when it rains and about 360 on a holiday eve. No one wrote that rule, the examples showed it.120360⁨52 past Fridays⁩⁨Next Friday⁩⁨110 to 130, sold 116⁩
39 ordinary8 rainy5 holiday evesBaked by the learned patternA big miss

Learned from 52 Fridays, the forecast says to bake about 120 on an ordinary Friday, about 100 when it rains and about 360 on a holiday eve. No one wrote that rule, the examples showed it.

Each dot is a past Friday and how many challahs it sold. The line is what would have been baked. Red marks a Friday baked far from what sold.

The model changes only when it is retrained

The forecast comes from a model trained on the examples, and it stays the same while it is used. A checked Friday becomes a new example, and only retraining brings it in.⁨The model⁩⁨fixed until⁩⁨the next⁩⁨training⁩⁨52 past⁩⁨Fridays⁩⁨Training⁩⁨Forecast:⁩⁨110 to 130⁩⁨Check: sold⁩⁨116⁩⁨Friday 53⁩⁨joins the⁩⁨examples⁩
The forecast comes from a model trained on the examples, and it stays the same while it is used. A checked Friday becomes a new example, and only retraining brings it in.

Definition

In machine learning the rule is not written by a person but found in examples. You show the system many past cases with their outcomes, for example past Fridays and how much bread sold, and it finds the pattern that links them. It then applies that pattern to a new case to predict a number or sort an item. Tom Mitchell's textbook Machine Learning (1997) defines learning broadly as any computer program that improves its performance at some task through experience. In a business the experience is examples, and the model gets better only when it is retrained on new ones.

Why it matters

Knowing how it works tells you what to check: a forecast is only as good as the examples behind it, and it can be wrong. Someone has to judge whether it is good enough before you act on it.

Directly related: AI, Model Training, AI Model.

Improvement tips

  • Decide before you start how often the forecast may be wrong and still be useful.
  • Give it examples that cover the cases you care about, such as holidays and bad weather, or it cannot learn them.
  • Retrain it when your business changes, such as a new product line or new opening hours, because it learned from the old situation.
  • Keep a person who knows the business in charge of acting on the forecast.

Common mistakes

  • Assuming the model improves by itself with use, when it improves only when it is retrained on new data.
  • Using machine learning for a decision you must explain to a customer, when you cannot explain the pattern it found.
  • Keeping a forecast in use after the business changed, when the old examples no longer describe it.
AI vs Machine Learning vs Generative AI

Related terms

AI and DataBeginner

AI

AI (artificial intelligence) is software that does tasks that used to need a person, such as writing, sorting or predicting, by learning patterns from data.

AI and DataIntermediate

Model Training

Model training is how an AI model learns: it sees many examples, checks its guesses against the right answers and adjusts itself until the errors shrink.

AI and DataIntermediate

AI Model

An AI model is the trained file at the heart of an AI tool: it takes an input, applies patterns learned from data, and returns a prediction, a label or text.

AI and DataIntermediate

Inference

Inference is the moment a trained AI model does its job: it takes a new question and produces an answer. It happens on every request, so it is the ongoing cost.

AI and DataBeginner

Generative AI

Generative AI is AI that creates new text, images, audio or code from a request. What it writes is a likely draft built from patterns, not a checked fact.

Quick check

What is the main difference between a rule and machine learning?

Choose an answer

Frequently asked questions

What is machine learning in simple words?
Software that learns a rule from examples instead of being told it. Show it many past cases with their outcomes, and it finds the pattern, then uses it to predict or sort new cases.
What does ML stand for?
ML is short for machine learning. It is one part of AI: the part where software learns from examples. Generative AI, the tools that write and draw, is built this way.
Is machine learning the same as AI?
No. AI is the umbrella for software that does tasks that used to need a person. Machine learning is the main way it gets there today: learning from examples. Generative AI is a newer part of machine learning.
When does machine learning make sense for a small business?
When you have many past examples of the same decision, the situation stays fairly stable, and a wrong guess now and then is acceptable. Forecasting stock or sorting incoming messages are typical cases.
Where do I already use machine learning?
Probably in several places: your email's spam filter, your card company's fraud alerts, "recommended for you" in online shops and quantity suggestions in some till and shop software. You use it without building anything.
Why are my machine learning forecasts wrong?
Usually because the examples do not cover the case: too little history, a holiday it never saw, or a business that changed since the data was collected. Check what it learned from before you blame the tool.
What drives the cost of machine learning?
Mostly the work around it, not the software: collecting and cleaning the examples, such as sales records with the dates and events that matter, and the time of someone who checks the results before you act on them.
When is machine learning the wrong tool?
When you have too little history, such as a few dozen rows, when the situation keeps changing, such as a new product line, or when you must explain each decision to a customer. A clear rule often does better there.
Do I need a statistics degree to use machine learning tools?
No, the tools do the mathematics. You still need someone who knows the business to judge whether a forecast is good enough before acting on it, and to notice when it starts to drift.
Does a machine learning model get smarter on its own?
Usually not. It uses the pattern it learned until someone retrains it on new examples. If your sales change and nobody retrains it, it keeps forecasting the old pattern.
Is machine learning safe to use with customer data?
It depends on what data you give it and on the vendor's terms, not on the method. Give it only the customer details the task needs, and check what the vendor does with them. Israel's Privacy Protection Law, 1981, as amended by Amendment 13, in force since 14 August 2025, governs collecting, keeping and using personal details, including details you paste into or upload to an AI tool. Educational material, not legal or tax advice. Check your own case with a lawyer or an accountant.
Can machine learning be wrong?
Yes. A forecast is a best guess from past examples, often given as a range. Decide in advance how wrong it may be and still be useful, and keep a person in charge of the decision.

Sources: Tom M. Mitchell, Machine Learning, McGraw Hill, 1997, section 1.1 "Well-posed learning problems" (cs.cmu.edu/~tom/mlbook.html), read 2026-09-25, Privacy Protection Authority, Amendment 13 questions and answers (gov.il/he/pages/tikun13_qa), read 2026-08-31

Last reviewed: 2026-09-25

I have worked with

What Clients Say

From different fields, at different stages of business

Dan Manto

Eclipse Capital

Real Estate Investment and Finance, USA

Arty McLabin

GameReady

Game Development Education and Outsourcing, International

Anna, Beautician

Anna

Beautician

After I finished a cosmetology course at a leading company, I was confused. I didn't know how to open a business, what to sell from home, or how to bring in clients. Since I reached out to Mobius, everything has changed. We built an organized plan with clear steps and no unnecessary risks. Today I see results, and I'm growing and developing every month.

Mark, Massage therapist, Gan Yavne

Mark

Massage therapist, Gan Yavne

When I finished a massage course, I imagined a different world. I found out it's a hard field, and financial stability isn't easy to find. When I came to Mobius, something changed. We started an organized process and built a business plan. Without Mobius? I probably would have given up. Today I believe in myself and in my own path.

Wondering where AI actually fits your business?

Alex helps Israeli businesses use AI and data where it pays off, and skip it where it does not. The first call is free.

+972 055-248-6151. Better on WhatsApp: I am in meetings most of the day and answer as soon as I am free.