An AI hallucination is a fabricated or unsupported claim that sounds plausible. In troubleshooting, treat AI output as a guide rather than established fact. section 1.3
Reduce the risk of accepting a hallucinated troubleshooting answer:
- Check whether the response is grounded in the details of the problem and the quality of the prompt. section 1.3
- Verify claims against reliable sources instead of relying on a plausible explanation. section 1.3
- Use follow-up prompts to expose gaps, clarify uncertainties, and continue the investigation. section 1.3
- Remove company details, private information, keys, and passwords before submitting code or troubleshooting data. section 1.3
AI-assisted troubleshooting
Sarah may paste code into an AI prompt while investigating an application problem. She should first remove sensitive details, then question the response, verify its claims with reliable sources, and refine the conversation when the initial answer does not explain the symptom. section 1.3
