Skip to content
●essay · 2026.06.20

How to get good at AI with free courses: a plan, not a list

by paul thomas·6 min·1,276 wordsESSAY

There is no shortage of free AI training. Search for it and you will find courses from Google, IBM, Anthropic, the Finnish government, and the UK government, most of them good for what they are, some of them not free at all. You could spend a month reading roundups of free AI courses without completing one.

The shortage is something else: a coherent sequence. A plan that turns content into capability.

This is that plan.

Why the course alone will not do it

Here is the honest version of what a free AI course gives you: a mental model and a vocabulary. That is not nothing. Understanding what a large language model actually does, why prompts matter, what "context" means inside a conversation, why the model hallucinates, what it is and is not well-suited for, these are useful things to know. A good course teaches them well in a few hours.

What a course does not give you is reps.

Skill comes from doing the thing, on your own work, with stakes. A swimming instructor can explain buoyancy in a classroom. That does not make you a swimmer. You become a swimmer by getting in the water, in a pool where falling short means getting wet rather than drowning, with someone who can spot what you are doing wrong.

AI skill follows the same pattern. Completing a course means you know the concepts. Using AI on a real task at your actual job, and then looking honestly at the output, is what turns those concepts into something you can repeat.

I called this "content is not capability" in my review of the UK government's AI Skills Hub. The Hub has indexed hundreds of free AI courses. It is a decent map. But the map is not the territory. The territory is your inbox, your client reports, your onboarding documents, your team meetings.

A plan that works, mostly for free

This is a four-step sequence. Each step is achievable in a week. None of it requires paying for anything until you reach step four, and even then only if it makes sense for you.

Step 1: Get the mental model right (a few hours, free)

Before you can use AI well, you need to understand what it is and what it is not. Two courses earn a clear recommendation here, both free including the certificate as of June 2026 (check the provider, because these things change).

Anthropic AI Fluency (anthropic.skilljar.com) is built around a four-part framework: delegation, description, discernment, and diligence. Those four words are more useful than they sound. Delegation is knowing which tasks to hand to AI and which to keep. Description is the craft of writing a prompt that actually gets you what you want. Discernment is reading the output critically. Diligence is the habit of checking it before you act on it. Complete the course and the final assessment, and you get a certificate at no cost.

Elements of AI: Introduction to AI (elementsofai.com) takes a slightly different angle, more technical grounding, less workflow-oriented. Free course, free certificate. If you want to understand what is actually happening inside a language model rather than just how to use it, this is where to go.

Either one, done properly, gives you a foundation that holds. Pick one this week and finish it.

Step 2: Pick one real task and do it with AI (this week)

This is the step most people skip, and it is the only one that actually matters.

Do not start with a made-up exercise. Start with something you were going to do anyway: a report you need to write, a document you need to summarise, a process you need to write up, an email thread you need to respond to. The task has to be real, from your actual job.

The AI Fluency skill guides on my free hub walk through the most common starting points in detail: how to use AI to summarise a long document, how to draft professional communications, how to build a standard operating procedure from scratch. Pick whichever matches something on your to-do list this week. Do it. Then read the output properly, fix what needs fixing, and notice what worked.

That last bit is the thing. Noticing what worked. Most people take the output, clean it up, and move on. If you pause for two minutes to ask yourself what the prompt did that got you a useful result, you are already ahead of most people using AI at work.

Step 3: Build the habit (one task a day for two weeks)

One task a week gives you ten tasks a year. One task a day gives you two weeks of daily practice in the first fortnight, which is enough to shift from conscious effort to something closer to instinct.

The rule is simple: each working day, pick one thing you were going to do without AI and do it with AI instead. A different task each time. Summarising one day, drafting the next, planning something the day after that. At the end of each attempt, spend two minutes on the same question: what worked?

You are not looking for perfection. You are looking for pattern. By the end of two weeks you will have a clear picture of where AI saves you time without much effort, where it needs a lot of steering, and where it is not worth it for your particular work. That picture is more useful than any course.

Step 4: Go deeper where it pays

At some point, if you are using AI seriously at work, free self-directed training stops being the constraint. The constraint becomes something else.

It might be that you need a more systematic framework than a course gives you. It might be that you are leading a team through AI adoption and you need help thinking through governance, how to set guardrails, how to build shared capability rather than individual pockets of it. It might be that you want structured feedback on your practice rather than just your own judgement.

That is the point where a structured programme, a coach, or specialist support starts to earn its cost. I offer training and enablement for teams working through exactly this: not the technology decisions, but the people side of AI adoption, how teams actually build AI capability rather than just access to AI tools. If that is where you are, get in touch.

For individuals still in the learning phase, the honest answer is that free will take you further than you expect. The gap between someone who has done a course and someone who has built a daily practice is large. Most people stay on the course side.

Where to start

The full list of genuinely free AI courses with certificates, plus the ones that claim to be free and are not, is in the complete free AI training guide. If you want the certificate angle specifically, here is the breakdown of which certificates are actually free.

The free hub at thehumanco.org/ai-resources has the AI Fluency course and the skill guides in one place, ungated. No email required. If you are going to follow this plan, that is where step one and step two live.

The courses are not the thing. The practice is the thing. The courses just give you enough to start.

Sources

// subscribe
One post like this a week.
Free. Unsubscribe in one click.