AI is not infallible
Generative AI has serious limitations every QA must know. Using it without judgment can introduce more risks than it solves. The golden rule: AI proposes, the human decides.
| Risk | Mitigation |
|---|---|
| Hallucinations (invents plausible data) | Always review and verify what is generated |
| Bias in training data | Cross-check with judgment and diverse cases |
| Privacy (leaking confidential data) | Don't paste sensitive or real customer data |
| Over-reliance | Keep the tester's critical thinking |
Never paste confidential requirements, credentials or real personal data into a public AI tool: you could leak protected information.
QuestionToca para voltear
What is an AI 'hallucination' and how is it mitigated?
Which is a good practice when using AI in QA?