Ask a room full of teachers what they want from AI training and the answer is remarkably consistent: less theory, more Tuesday afternoon. Nobody there needs a lecture on attention mechanisms. They need to know what to do when a Year 9 student hands in a history essay that reads like a press release, and they need to know before first period.
That gap between the technical and the usable is where the best generative AI for education courses live. The weak ones stop at “AI can help you plan lessons.” The good ones get you from a blank page to a differentiated worksheet, a marking rubric and a set of feedback comments in under twenty minutes, and they leave you knowing which parts to trust and which to rewrite.
What these courses actually teach
A well-built course spends most of its time on four things: prompt design, output judgement, workflow integration, and the limits of the tool. Prompt design is the easy part to sell. Output judgement is the part that matters, because a lesson plan that looks confident and reads fluently can still be pitched at the wrong reading age or quietly invent a historical event.
Expect to work on concrete tasks rather than abstractions:
- Rewriting a single worksheet at three reading levels for a mixed-ability class
- Generating a marking rubric that matches your school’s existing assessment language
- Building a question bank for retrieval practice that avoids the same three question stems
- Drafting feedback comments a student can act on, not just read
- Writing parent-facing communication about a sensitive topic without sounding like a chatbot
- Spotting AI-generated student work without relying on unreliable detector tools
Keep an eye out for any plagiarism policy content. A course that never mentions academic integrity, data protection or what you should and shouldn’t paste into a chatbot is only telling you half the story.
Three types of course, and one is usually a better bet
Courses about AI
These explain how large language models work, cover training data and hallucinations, and often touch on ethics. Useful context if you want to lead a department or write policy, but they rarely change what you do in a classroom on Monday. If that’s genuinely what you want, a broader generative AI applications course will cover the mechanics in more depth than most education-branded versions.
Courses about teaching with AI
This is the category most educators actually need. The focus sits on planning, assessment, differentiation, feedback and admin, with the tool treated as a colleague who drafts fast and checks nothing. Good versions are built by people who still teach.
Courses about governance
Aimed at headteachers, IT leads and safeguarding staff. Expect policy templates, risk registers, procurement questions and staff training plans. UNESCO’s 2023 guidance on generative AI in education and research is a common reference point, and the UK Department for Education published its own position paper the same year.
What a realistic six-week syllabus looks like
If a course promises mastery in ninety minutes, it’s selling a demo, not training. A credible structure tends to run something like this:
Week 1: What the tools do and don’t do. Hands-on with two or three models side by side, comparing outputs on the same prompt. Week 2: Prompt structure, including how to give a model your context, constraints and audience. Week 3: Lesson and unit planning, with an emphasis on adapting rather than accepting. Week 4: Assessment, rubrics and feedback. Week 5: Differentiation and support for students with additional needs. Week 6: Policy, ethics, bias and a personal workflow you’ll actually keep using.
Somewhere in there you should be producing artefacts you can use: a real lesson resource, a real rubric, a real policy draft. If the assessment is a multiple-choice quiz, the course is about awareness, not practice.
Free versus paid: where the money actually goes
Plenty of excellent material costs nothing. Free AI courses for teachers cover much of the same ground as paid entry-level programmes, and providers like Google, Microsoft and Common Sense Education publish educator-facing training without a paywall. What you tend to pay for is structure, a cohort, feedback on your work and a certificate someone recognises.
Before spending anything, check what genuinely free certification options exist in the space. This breakdown of free AI certificate programmes is honest about what the $0 tier includes and where the upsell starts. If your school is funding the training, the calculations change, and it’s worth reading a comparison of AI certification courses to see which price points buy real feedback rather than a video library.
Six things to check before you enrol
- Date of last update. Anything recorded before mid-2024 is describing a different generation of tools.
- Which products are used. A course built entirely around one vendor’s chatbot is a sales pipeline unless it teaches transferable prompting.
- Who wrote it. Look for a named educator with a school, not a job title at a training company.
- Assessment format. Artefacts beat quizzes every time.
- Data and safeguarding content. Non-negotiable if you work with minors.
- Refund window. Two weeks is reasonable. No refunds plus a big discount timer is not.
Red flags that keep appearing
Watch for courses that promise to “future-proof” your career, show only polished output with no discussion of hallucinations, or casually suggest uploading student data into a consumer chatbot. Another tell: modules titled with tool names that have since been renamed or discontinued. The education AI space moves fast enough that a course’s half-life is roughly eighteen months, so a provider who isn’t publishing updates isn’t maintaining the product.
One more, subtler flag. If every example involves a generic primary school and no specific subject, year group or curriculum framework, the content was probably written for a general corporate audience and relabelled. Maths teachers can tell within a page.
The test that separates good courses from expensive ones
Here’s a practical filter you can apply in about ten minutes. Read the course outline and ask a single question: after this module, what will I be able to do that I couldn’t do before?
If the answer is a verb plus a task, the course is probably worth your evening. “Build a three-tier question set for a Y10 geography unit on glaciation and adapt it for a student reading two years below age-related expectations.” That’s a real skill. If the answer is a noun, “understand the landscape of AI in education,” you’re buying vocabulary, and you can pick that up from a forty-minute podcast.
Also worth doing: pick one lesson you’re teaching next week and run it through whatever free trial the course offers. If the course’s method makes that lesson noticeably easier to prepare or easier for students to access, you’ve got your answer. If you finish the trial with a warm feeling and no new resource in your folder, you’ve got a different answer.
The best generative AI for education course is rarely the most expensive one or the one with the longest syllabus. It’s the one that leaves you with a working habit, a sensible scepticism about confident-sounding output, and at least one thing you can hand to a class before the week is out.