AI in Testing for Quality Assurance (QA)s
AI in testing covers two things: using AI assistants to draft test cases, generate test data and review test code, and testing products that contain AI, whose output is non-deterministic. Both need human judgement, because generated tests can look plausible while checking the wrong behaviour.
Stage: CI/CD and Modern QA · Part of the Quality Assurance (QA) roadmap · Last reviewed 19 November 2026
What to learn in AI in Testing
- Prompt an assistant to draft test cases from a requirement and then review them
- Check generated automation code for wrong assertions and missing cases
- Generate realistic synthetic test data without using personal data
- Define evaluation criteria for an LLM feature's answers
- Build a small golden set of prompts and expected qualities
- Probe an AI feature for hallucination, bias and prompt injection
AI in Testing interview questions
How do you test a feature whose output varies on every run?
Define pass criteria as properties instead of exact strings: correct facts, required format, safety, tone and length. Run a fixed evaluation set many times, score the results with rules or a reviewer, and track the pass rate over releases.
What risks come with AI-generated test cases?
They may repeat the requirement without finding edge cases, assert incorrect expected results, invent features, or leak confidential data into a prompt. Treat the output as a draft and review each case against the real specification.
Where AI in Testing fits in the Quality Assurance (QA) roadmap
AI in Testing belongs to the CI/CD and Modern QA stage (intermediate level), which takes about 5 hours. Quality in the pipeline.
Other units in this stage
Back to the full Quality Assurance (QA) roadmap