How we use AI
Last updated 2026-09-26
We use AI tools to make parts of the Programme, and we want you to know exactly which parts, how they were checked, and what was not checked.
1. What is made with AI
| Part of the Programme | How AI was used |
|---|---|
| Course notes, workbooks and exams | Drafted with the help of an AI model (Claude, made by Anthropic) |
| Lab and exercise solutions | Drafted with AI help; the code is run by automated tests |
| The glossary (about 3,500 terms) | Drafted with AI help |
| Synaptova explainer and deep-dive videos | Narrated by an AI-generated voice (the Kokoro text-to-speech model) and animated with generated visuals. These videos are labelled. |
2. How it was checked, and what was not
- Checked: the code in lab and exercise solutions is executed by automated tests, and a solution that fails the tests is not published.
- Not checked: the notes, workbooks, exams, glossary and video scripts have not been reviewed line by line by an independent human subject-matter expert. Passing a test shows that code runs and gives the expected result in the cases we tested. It does not show that an explanation is complete, current, or the best way to do something.
- We are responsible for what we publish. We do not describe the material as "expert-reviewed" or "industry-approved".
3. What is not AI
- Recommended lectures from other creators are shown through the owners' official players. We did not make them and do not change them. See Credits and Licences.
- Ask Nova, the AI tutor in the app, is an AI model. When you ask Nova something, we send your message, the recent messages of that conversation, your first name, the lesson you are on and the lesson objectives you recently got wrong to Cloudflare Workers AI, which runs the model and returns the reply. Nova is switched off while an exam is running, and it will not give answers to course quiz, exam or workbook questions. Its replies can be wrong: check them like any other AI output.
- Nothing else in the app calls an AI service. Python labs run inside your own browser on your own device.
- We do not use your data to train AI models. Do not put personal data (yours or anyone else's) into your messages to Nova. See the Privacy Policy.
4. How to treat AI-generated material
- Verify before you rely. Check important facts, library behaviour and commands against the official documentation, especially anything you will use at work.
- Run the code yourself. If a result surprises you, trust the documentation and your own tests over our text, and tell us.
- Do not treat it as advice. The Programme is general education. It is not professional, legal, financial or career advice.
- Expect that tools change. Libraries, model names and prices move quickly. A note can be out of date even when it was right when written.
- Your own use of AI tools. Quizzes, video checks and exams test your own understanding: answer them yourself, without AI tools or help from other people. Nova is switched off during exams for this reason. In notes, workbooks and labs you may use Nova or other AI tools to explain ideas and error messages, but write and run the code yourself. Using AI or anyone else to answer an assessment can lead to the attempt not counting and, if repeated, to your certificate being refused.
5. Found a mistake?
Please tell us. Email [email protected] with the lesson ID (for example M2-L05), the exact sentence or step, and what you think is right. We acknowledge reports within 48 hours and correct confirmed errors in the lesson. If a mistake affects your work or assessment result, we will review your result. If you are not happy with the reply, use the online grievance form.
6. Labels and honesty
We label AI-narrated videos. Illustrative profiles, scores and dashboards in marketing material are examples and are marked as such. If we change how we use AI in a way that matters to you, we will update this page. For the legal terms, see Terms section 6 and the Disclaimer.