Start with the fundamentals, improve your practice, or learn to build.
Start using ChatGPT, then practice on a task you care about.
Start a conversation, choose a useful first task, and learn when to move from a quick chat to a larger task in Work. Includes voice basics and a route into the AI Foundations course.
Good for Your first conversations with ChatGPT
No programming required.
Reusable prompts with guidance for testing and adapting them. Includes an idea-challenging “Devil’s Advocate” example and a practical checklist for evaluating outputs.
Good for Improving a prompt you actually use
Test prompts with your own tasks and model. Reusing the library’s prompts requires attribution to Ethan and Lilach Mollick under CC BY 4.0.
Build your understanding or develop something of your own.
Structured learning, with costs and prerequisites up front.
An introduction to what AI is, what it can do, and how to think about its effects. A useful foundation before focusing on a particular tool.
Good for Understanding AI beyond a single product
An account is required for complete course access. Six chapters with exercises; check the course FAQ for completion requirements.
Hands-on practice in prompting, workplace tasks, and responsible use. A structured option for people who want a short course with a completion certificate.
Good for A structured introduction to AI at work
No technical experience required. Google lists US/Canada pricing at US$49 per month after a 7-day trial; check local enrollment pricing.
Google completion certificate; not a professional license.
A beginner learning path with seven modules covering core AI concepts, generative AI and agents, language, speech, vision, information extraction, and retrieval-augmented generation. Assumes basic computing and math knowledge.
Good for Technical beginners who want a guided AI overview
Basic computing and math knowledge; aimed at developers and AI engineers.
Code, APIs, and agents. Programming experience is expected.
A 21-lesson curriculum with Python and TypeScript examples: chat, search, images, retrieval, agents, and application security.
Good for Building a foundation through working examples
Python or TypeScript fundamentals. Choose a supported model service or local setup; the course now directs GitHub Models exercises to Microsoft Foundry Models. Cloud usage may cost extra.
Practical examples and guides for building with the OpenAI API. Start with a use case and work through the accompanying code.
Good for Finding an implementation pattern
Coding experience; examples are mostly Python. API examples require an OpenAI account, API key and any billed model usage.
Learn agent concepts, work with established frameworks, and build an evaluated final project. Includes retrieval, observability, and a game-playing example.
Good for Moving from chat prompts to working agents
Basic Python and LLM knowledge; a Hugging Face account for publishing projects.
Free course certificates: a fundamentals certificate after Unit 1, and a completion certificate after the required use case and final challenge.
Courses on language models, agents, context engineering, audio, and other practical topics in the Hugging Face ecosystem. Check each course for prerequisites.
Good for Exploring additional technical subjects
A prompt caching guide to shared prefixes, stable tool definitions and cache breakpoints. Includes diagnostics for misses; compare cache-write costs with the savings from repeated reads.
Good for Reducing avoidable latency and input costs
An API integration with repeated context. Rates and cache behavior vary by model; a persistent session does not guarantee a cache hit.
A practical guide to balancing model capability, latency and cost, with advice on instructions, long-running work and production checks. Test vendor recommendations on your own tasks.
Good for Choosing a model for a real workload
Basic API or coding-agent experience. Compare cost per successful task, not token price alone; benchmark before changing a production workflow.
Calls for labs to stop testing advanced mathematics on inaccessible proprietary models, alongside recommendations for attribution, proof artifacts and community-led human understanding when results are released.
Good for Evaluating how AI research claims are shared
Mathematical research context. These are scholarly recommendations, not verification of any particular proof or endorsement of a lab.
No matches. Try another search or type.
“Checked” means we reviewed the linked source, not that we tested every tool or completed every course. Availability and prices can change. Completion certificates are labeled separately from professional qualifications.
Allow Google Analytics to measure visits and link clicks? It’s optional. Privacy details