Teachers are being asked to use AI before anyone has explained it to them. The better courses fix that in a few hours. Google’s Generative AI for Educators with Gemini is self-paced, needs no technical background, and ends in a certificate you can submit to your district toward professional development credit, where district and state rules allow it. Anthropic runs a free AI Fluency course for faculty, instructional designers and educational leaders. The Raspberry Pi Foundation offers a free course built with Google DeepMind on understanding AI. Check the current terms on each provider’s page before enrolling.
This guide covers what each teaches, how they differ in philosophy, and the classroom decisions no course makes for you.
What You Will Learn
Foundations before tools
The strongest courses start with the concept, not the button. Google’s course promises a grounding in AI: a definition, the openings and constraints this technology brings, and the habits of responsible use. The Raspberry Pi course similarly covers a definition of AI, the way such systems get built, the range of problem-solving methods behind them, and effective, ethical use of the tools available today.
Practical classroom tasks
Google is explicit about the daily payoff. You learn to use tools like Gemini and NotebookLM to cut the time spent on routine correspondence, tailor teaching to varied learning styles and abilities, and add creative depth to lessons and activities.
A framework rather than a tool list
Anthropic’s course is built around a named model. It builds on the 4D AI Fluency Framework taught in the provider’s core course and covers putting AI fluency to work in the way courses and learning are designed. Frameworks travel better than tool tutorials, which age within a year.
Language discipline
One design choice from the Raspberry Pi course deserves wider adoption. Its authors never anthropomorphise AI systems, replacing phrases such as ’the model understands’ with ’the model analyses’, and they avoid using AI as a singular noun, preferring ‘AI tool’ or ‘AI system’. Teaching that vocabulary to students is arguably more valuable than teaching them any particular product.
Course Structure and Format
Self-paced and short
Google’s course is self-paced with flexible scheduling and no technical experience required, which makes it realistic during a term rather than only in the summer.
Cohort-free but assessed
Anthropic’s route ends with an assessment: after finishing, you sit a closing assessment, after which a completion certificate is issued. Its curriculum moves from an introduction and framework review through applications for educators to applying fluency in course design.
Built for mixed confidence levels
The Raspberry Pi course was deliberately designed to put educators of widely differing confidence on equal footing, aiming to bring every teacher to the point where they hold a confident classroom conversation about AI.
Comparing the three
| Course | Provider | Cost | Focus |
|---|---|---|---|
| Generative AI for Educators | Free to start, certificate for PD credit | Everyday classroom tasks with Gemini | |
| AI Fluency for Educators | Anthropic | Free | Framework, course design, institutional strategy |
| Understanding AI for educators | Raspberry Pi Foundation with Google DeepMind | Free | Conceptual grounding and accurate language |
Reading that table
If you want tomorrow’s lesson planning to get faster, take the Google course. If you are redesigning a syllabus or advising a department, take the Anthropic route. If you will be teaching students about AI itself, take the Raspberry Pi course, because its terminology discipline is the point.
Product, Course, App and Platform Experience
Vendor platforms
Google’s course sits inside a wider training catalogue and is packaged for busy teachers: self-paced, flexible, and immediately applicable to your workflow. Expect polished production and a clear product bias toward Gemini and NotebookLM, which is fine as long as you know it going in.
Academy platforms
Anthropic delivers through a course platform with registration, saved progress and a final assessment. The material was built in partnership with academic experts, Prof. Joseph Feller of University College Cork and Prof. Rick Dakan of Ringling College, which shows in the framing.
Non-profit platforms
The Raspberry Pi Foundation publishes its course as part of a broader educator programme, alongside free classroom resources. There is no upsell, and the pedagogy is the product.
What to check before you commit an afternoon
Three things. Does the certificate count for your district’s PD requirements? Google notes that PD credit depends on district and state requirements, so confirm locally. Does the course use tools your school actually permits? And does it teach concepts you can reuse when the tool changes?
Benefits of AI in Education
Time returned to teaching
Drafting, summarising and differentiating materials are the fastest wins. Google frames the outcome as working smarter rather than harder by treating AI as a collaboration tool.
Genuine differentiation
Personalising a worksheet for three reading levels used to cost an evening. That is the clearest classroom application, and it is one Google’s course targets directly through tailoring teaching to varied learning styles and abilities.
Confidence in conversations
Parents, students and administrators all have questions. A course that gives you a solid conceptual foundation so you can ask the right questions and form your own perspective is worth more than any prompt library.
A worked classroom example
Take a teacher with five classes preparing a unit on ecosystems. Generating three differentiated reading passages and a set of comprehension questions might take ninety minutes by hand and twenty-five minutes with an AI assistant plus review. That is roughly an hour returned per unit. Across a term with eight units, that is a working day. The catch: every generated passage needs a factual check, because a fluent paragraph about food webs can still contain a wrong claim, and students will quote it back to you.
Where AI does not belong
Some tasks should stay untouched. Final grades on high-stakes work, safeguarding conversations, references for students, and anything requiring professional judgement about a child. Delegating those erodes exactly the trust that makes teaching work.
A simple three-question test
Before using a tool on any task, ask three things. Can I verify the output quickly? Does any student’s personal information leave my control? Would I be comfortable telling parents I used it here? A no to any of these means do it yourself.
Real-World Applications and Case Studies
Planning and materials
Lesson outlines, rubrics, differentiated texts, quiz banks. Low risk, immediately useful, and easy to review before use.
Feedback and communication
Drafting parent emails and first-pass comments on student work. Google names drafting correspondence as a core use. The human judgement stays with the grade; only the phrasing is assisted.
Teaching about AI
For computing and digital literacy lessons, the subject becomes AI itself. The Raspberry Pi course exists precisely to help educators bring AI into their classroom confidently, and its language rules are directly transferable to students.
Institutional strategy
Department heads and instructional designers face a different problem: policy, assessment integrity and staff training. Anthropic’s course targets exactly that group, covering teaching practice and institutional strategy.
The department rollout pattern
Individual training rarely changes a school. What works is narrower: one department, one term, one agreed task. Pick a task every teacher in the department already does, such as writing differentiated reading passages or generating retrieval-practice questions. Agree a shared prompt, a shared verification rule, and a shared way of noting when AI was used. Compare workload notes at the end of term.
Two things usually surface. First, the time saved is real but smaller than the enthusiasm suggested, because review takes longer than people expect. Second, the quality gain shows up in differentiation rather than speed, since producing four versions of a text was previously impossible rather than merely slow.
Students notice more than adults assume
Learners can usually tell when a worksheet was generated and not read. Explaining that you used a tool, checked it, and changed three things models exactly the behaviour you want from them. It also removes the awkwardness of pretending otherwise.
Ethical Considerations and Challenges
Accuracy is your responsibility
Anything a model produces for a classroom is a draft until a teacher verifies it. Build the check into your routine rather than trusting a good first impression.
Student data
Never paste identifiable student information into a general-purpose tool. Settle this with your school before you experiment, not after.
Assessment integrity
If students can generate an assignment in thirty seconds, the assignment measures the tool rather than the learner. Redesigning tasks is more effective than trying to detect generated text.
Equity of access
Not every student has the same access at home. Any AI-dependent homework widens gaps quietly.
Language shapes belief
This is why the Raspberry Pi approach matters. Describing a system as understanding rather than analysing teaches children something false about how the technology works.
What to Know Before Deciding
Free is the norm here
All three of these are free to take. That removes the usual excuse and also means you can sample one without a business case.
Short does not mean shallow
A few hours of well-designed material beats a semester of unfinished reading. Pick the shortest course that covers concepts, not just clicks.
Certificates vary in usefulness
Google offers a certificate you can present for professional development credit, subject to district and state rules; Anthropic issues a completion certificate once the closing assessment is passed. Confirm what your employer recognises.
Tool-specific content ages fastest
Product walkthroughs date within a year. Frameworks, terminology and ethics do not. Weight your choice accordingly.
One course will not settle school policy
Training builds capability. Policy still needs a decision from leadership about permitted tools, data rules and assessment design.
Frequently asked questions
Do I need technical experience?
Are these courses free?
Will I get a certificate?
Which should a total beginner take first?
Next Steps and Resources
Pick one task, not one tool
Choose the job you most resent: differentiating texts, writing report comments, building quiz banks. Take whichever course addresses that job first.
Do it in two sittings
These courses are short enough to finish in a week of evenings. Book the time before you enrol, because term time expands to fill any gap.
Write your own classroom rules
One page: which tools are permitted, what student data never goes near them, how you verify factual content, and how you will tell students what you used. That page is what turns a certificate into practice.
Share it with one colleague
Departments change faster than individuals. Running the same course with one other teacher, then comparing notes, is the cheapest professional development available.
Keep practising after the certificate
Skills fade if they never meet a real deadline. For ongoing, applied practice alongside a formal course, explore Coursiv AI lessons and use them as the weekly habit under your training.