Module 3 Basics of an AI assistant
Hours 40–60, 20 hours, the same for every track. Signed off by the mentor with a walkthrough. Reviewed at Gate 0.
The full training content, lesson by lesson, is in Basics of an AI assistant: training.
Purpose
An AI assistant can explain code, suggest a plan, and draft tests in seconds. It can also be confidently wrong. Module 3 teaches each person to use one well: to ask clearly, to check every answer, and to keep patient data, personal data, and secrets out of it (Principle 13 and Principle 15).
The default assistant is Google Gemini AI Mode. A person may use another AI assistant, with their mentor's agreement and within the organisation's AI use policy (Principle 9).
Outcomes
- Learning outcome 17: use an AI assistant to get training advice, plan continuing professional development, compare concepts, explain code, and convert between user stories, Given-When-Then scenarios, and Selenium JavaScript, checking every answer.
Depth by track
| Module | Band 3 | Band 4 quality assurance | Band 4 test engineering | Band 5 quality assurance | Band 6 quality assurance | Band 6 test engineering | Band 7 test engineering | Band 7 test management |
|---|---|---|---|---|---|---|---|---|
| Module 3 Basics of an AI assistant | Independent | Independent | Independent | Independent | Independent | Independent | Independent | Independent |
Session plan
| # | Session | Duration | Format | Who |
|---|---|---|---|---|
| 1 | Hours 40–45: what an AI assistant is and is not; the organisation's AI use policy; what never goes into a prompt | 2 hours | Core | Cohort |
| 2 | Hours 40–45: prompting: context, goal, format, and examples; follow-up prompts | 3 hours | Core | Cohort |
| 3 | Hours 45–52.5: asking for training advice, and checking it against the programme | 1.5 hours | Practice | Each person |
| 4 | Hours 45–52.5: planning your own continuing professional development with an assistant | 2 hours | Practice | Each person |
| 5 | Hours 45–52.5: comparing and contrasting concepts, such as manual and automated regression testing | 1.5 hours | Practice | Pairs |
| 6 | Hours 45–52.5: explaining source code, checked against the code: your Module 1 function and Module 2 script | 1.5 hours | Practice | Each person |
| 7 | Hours 45–52.5: check-in with the mentor | 1 hour | One to one | Person, mentor |
| 8 | Hours 52.5–60: converting a user story into Given-When-Then (Gherkin) scenarios | 2 hours | Practice | Each person |
| 9 | Hours 52.5–60: converting the scenarios into Selenium JavaScript, running it, and converting back | 3 hours | Practice | Each person |
| 10 | Hours 52.5–60: rehearse, then the walkthrough to the mentor | 2.5 hours | One to one | Person, mentor |
Each person's sessions add up to 20 hours.
Activities
- Read the organisation's AI use policy. Never put real patient data, personal data, credentials, or confidential code into a prompt. Use the fixture site and the practice repository instead.
- Ask for training advice: for example, "I am a manual tester at Band 5 learning Selenium with JavaScript. What should I practise in the next 20 hours?" Compare the answer with this programme, and note where it is wrong or too general.
- Draft your continuing professional development plan for the next year with the assistant. Edit it until it is yours. Bring it to your individual learning plan meeting at Gate 0.
- Ask the assistant to compare and contrast two concepts, such as a test plan and a test strategy, or manual and automated regression testing. Check its claims against a source you trust.
- Paste your Module 2 script and ask the assistant to explain it. Mark each sentence of the explanation as right or wrong.
- Take a user story, such as "As a patient, I want to book an appointment online, so that I do not have to phone." Ask for Given-When-Then scenarios. Correct them. Ask for Selenium JavaScript for one scenario against the fixture site. Run it, fix it, and then ask the assistant to turn the code back into a scenario and a user story.
Evidence
Evidence 3: a walkthrough to the mentor of the person's own sessions with Google Gemini AI Mode, or the assistant the mentor agreed, on their own system, showing:
- clear prompts, and follow-up prompts that improve an answer
- asking for training advice
- a draft plan for the person's own continuing professional development
- comparing and contrasting two concepts
- explaining a piece of source code, checked against the code itself
- converting a user story into Given-When-Then (Gherkin) scenarios, those into Selenium JavaScript, and back, with every output checked and corrected by the person.
Assessment
The mentor signs off the walkthrough when the person can do each step on their own system, show where they checked and corrected the assistant, and has kept every prompt free of real patient data, personal data, and secrets. Gate 0 (hour 88) records the sign-off.
| Walkthrough sign-off | Entry |
|---|---|
| Person | |
| AI assistant | |
| Prompting and follow-up | Yes / No |
| Training advice, and a continuing professional development plan | Yes / No |
| Compare and contrast; explain source code | Yes / No |
| User story to Given-When-Then to Selenium JavaScript, and back | Yes / No |
| Every answer checked; no patient data, personal data, or secrets | Yes / No |
| Mentor, and date |
Resources
- Google Gemini AI Mode, or the assistant agreed with the mentor.
- The organisation's AI use policy.
- Your Module 1 function, your Module 2 script, and the fixture site: https://testingexamples.github.io/en-001/practice/
View this document on GitHub materials/modules/module-3-basics-of-an-ai-assistant/index.md