Answer in brief
Choose an AI course by the task you want to be able to do in a month, not by the length of the programme. Work out which of three kinds you need: for users, for one profession or for engineers. Read the modules for verbs you can check and for review of your own work, take a free course from a model maker first, and read the contract and the refund clause before paying.
Name the task for the month before you open a catalogue
Most people who ask how to choose an AI course start from the catalogue: which school, how many weeks, what price. The useful starting point is one sentence about yourself. In a month I want to answer customer emails in half the time. The United States Federal Trade Commission (FTC) puts the same question near the top of its guide to choosing a vocational programme: think first about whether you need more training for the job you want.
A task sorts the market at once. It tells you which tool you will work in, what level you start from and which modules you can skip. It also gives you a test for the end: either the emails go out faster or they do not. Without a task the choice falls to the longest programme, the loudest page or the biggest discount.
Three kinds of AI courses and how to tell which one a page sells
The same words on a sales page hide three different products. The first kind is for users: people who work with ready assistants and need to frame a task, write a clear prompt and check the answer. The second is for one profession: a designer, a marketer or an accountant learns where AI fits inside a trade they already know. The third is for engineers: code, programming interfaces, data and systems that run without a person watching. The entry requirements usually give the kind away: a course that asks for a programming language is not for beginners.
The companies that make the models sort their own learners in the same way. OpenAI Academy lists four paths on its front page: for knowledge workers, for builders, for leaders and for education. Grow with Google separates AI Essentials, for people new to the subject, from training for educators, small businesses and students. Anthropic’s Claude Academy keeps its general courses apart from the guides to Claude Code, for building software, and Claude Platform, for developers.
Reading the programme: verbs you can check and work that someone reviews
A programme is a list of promises, so read its verbs. Get acquainted with, learn about, overview of: these cannot be checked, by you or by the school. Write, compare, verify, set up, automate: these end in something that exists and can be shown. Go through the modules and mark each one with what you will hold at the end of it.
Two public frameworks show what a full programme covers. UNESCO’s AI competency framework for students, published in 2024, describes 12 competencies across four dimensions (a human-centred mindset, ethics of AI, AI techniques and applications, AI system design) and three levels: understand, apply, create. It was written for school curricula, but the grid suits an adult too: ask which level a course takes you to. Anthropic’s AI Fluency course names four competencies: delegation, description, discernment and diligence. Only one of them is about wording a request; the others are about deciding what to hand over, judging the result and answering for it.
Then look for your own task in the practice. A course where everyone repeats the teacher’s example checks attention, not skill. Ask what you will do with your own material, who reviews it, by what criteria and how soon. If the page is silent about review, ask for a sample of feedback before paying.
Promises to distrust: a guaranteed job, a profession in two weeks, a certificate
The FTC says it without softening: no school can guarantee you a job when you graduate, and a school is not an employment agency. Its guide adds why such promises appear: some schools promise more than they can deliver in order to raise enrolment and profit. The guide concerns American vocational schools, but the mechanism is the same anywhere. A guaranteed income after an AI course is the same promise wearing a different number.
A profession in two weeks fails a simple comparison. The model makers measure their own introductory courses in hours: Claude Academy lists AI Fluency at about four hours and its course on AI capabilities and limitations at about three and a half, and Grow with Google lists AI Essentials at five. Their titles and descriptions speak of foundations and fundamentals, not of a profession. A certificate deserves the same calm: Claude Academy gives a completion badge for a course, and the FTC advises calling employers to ask what training and experience they would find helpful.
Refunds are set by law and by the contract, and both differ by country. In Russia, article 32 of the consumer protection law lets a consumer cancel a service contract at any time, on condition of paying the provider the expenses actually incurred. The European Union’s Your Europe portal describes a 14-day right to cancel a service bought online, counted from the day the contract was agreed, with exceptions such as online digital content you started to download or stream after expressly agreeing to lose that right. Read the refund clause before paying, and for a large sum check your case with a lawyer.
What the model makers themselves teach free of charge
Before paying anyone, spend an evening with the makers. Claude Academy describes itself as free courses, tutorials and use cases from Anthropic’s education team; its course on AI capabilities and limitations, by its own page, assumes no technical background. Grow with Google says in its questions section that it offers workshops at no cost, among them ones for educators, business owners and students. OpenAI Academy invites visitors to join; its front page does not state a price, so read the terms when you sign up.
These materials have limits. Each catalogue leans toward the maker’s own products, the pages named here are in English, and nobody tells you where your prompt went wrong. Catalogues also change, so check the pages as they are today. What a few free hours do give is vocabulary and a yardstick: afterwards you can see which modules of a paid programme repeat what you already know and which add practice, review and structure.
Checking a school before you pay: a trial lesson, the author’s work, the contract
Start with what the school gives away. A free lesson or an open module shows how the teacher explains, at what pace and with what examples. Then look at the author: not how many followers they have, but what they have made with these tools and whether you can open it. The FTC adds a rule about paper: read the written materials, including any contracts, before you settle on a programme, and if a school will not give you documents to review before you commit, do not enrol.
The same guide suggests asking what share of students complete the programme, whether you can talk to recent graduates, and whether you can cancel within a few days of signing up. Local law may set the form as well: Russia’s law on education says in article 54 that an education contract is made in simple written form and states the full cost of paid services, and that its details must match what the organisation’s official site says on the day of signing. Whatever the country, save the programme page on the day you pay.
Getting the most from any course: one project, a fixed schedule, your own notes
The course you choose matters less than how you take it. Bring one project and carry it through every module: the customer emails, the product descriptions, the weekly report. Each new technique is tried on the same material, so by the end you have a working process and not a folder of exercises. Put the hours in the calendar as you would a meeting with a client.
Keep notes that you will actually reuse. Write down the prompts that worked, with the task, the context you gave and the check you ran on the answer. Write down the failures too: where the model invented a fact, lost the thread or ignored an instruction. A month later these notes are worth more than the recordings.
How the VITON13 SCHOOL course is built and whom it suits
Some of this needs no course at all. This week you can write your one-sentence task, take a free course from a model maker, and write five prompts for your own material, checking every answer against a source you trust. If that already solves the task, stop there. A structured course earns its place when the task is larger than a prompt and you need an order of steps, practice on your own work and a route to a process that runs without you.
The Artificial Intelligence course at VITON13 SCHOOL is an online course with practice on the learner’s own task. It is built for that case: it goes from framing a task through prompting and verification to automating your own processes and building agents. Its seven modules are Working with AI deliberately, Prompting as engineering, Research and verification, Automating processes, AI agents, Your own systems, and Safety and responsibility. Judge it by the rules above: match the modules to your sentence, read the current terms on the course page, and try the free templates in the school library first.
Practical checklist
- Write one sentence: what you want to be able to do in a month without help.
- Decide which kind you need: for users, for your profession or for engineers.
- Mark every module of the programme with the thing you will hold at the end.
- Take one free course from a model maker before comparing paid ones.
- Read the contract and the refund clause, and save the programme page on the day you pay.
Questions and answers
Are the free courses from model makers enough to start with AI?
For a first step, yes. Anthropic and Google list their beginner courses at a few hours each, and OpenAI runs an academy of its own. This material gives vocabulary and basic habits, but it leans toward each company’s products and nobody reviews your work. It suits a start and a test of interest more than a finished skill.
How long does it take to learn to work with neural networks?
It depends on the task, not on a school’s calendar. An introductory course from a model maker takes an evening or two. Turning that into a habit at work takes weeks of regular practice on one real project. Be wary of any fixed term attached to the word profession.
Does a certificate from an AI course help to get a job?
A certificate shows that you attended and finished. The FTC’s guide suggests asking employers directly what training and experience they value. In practice a portfolio of checked work on real tasks says more than a document, so choose a course by the work it leaves you with.
Can I get my money back if an AI course does not suit me?
That depends on the country and the contract. Russian consumer law allows cancelling a service contract at any time while paying the provider’s actual expenses; the EU describes a 14-day right for services bought online, with exceptions. Find the refund clause in the contract and ask a lawyer about your own situation.
Do I need programming skills to take a course on AI?
Not for a course aimed at users or at one profession: these work with ready assistants and plain language. Programming appears in courses for engineers, where the final work is code. The entry requirements on the page should say which kind it is; if they are silent, ask before you pay.