What you will be able to do
Use a simple source-checking routine to decide whether an AI answer is safe enough to use.
Estimated time: 25 minutes
Plain-language summary
AI can produce confident wording that is incomplete or wrong. For important work, trace each claim back to a trusted source, check the context and keep a human responsible for the final decision.
Guided example
Imagine you need a short summary of an approved workplace policy. Open the policy yourself and use this original prompt template:
Using only the source text I provide, summarise the policy in five plain-language bullet points. For every point, name the source heading that supports it. Mark anything not supported by the source as "not found". Do not invent an answer.
After the AI responds, compare each point with the original policy. Correct or remove anything that the source does not support.
Practice
Choose a low-risk, non-confidential source such as a public organisation page. Ask an AI tool for three claims, then mark each claim as supported, partly supported or not found.
Feedback or model response
Logbook prompt
What made one claim trustworthy or untrustworthy, and what did you change after checking the source?
Next step
Use the checklist the next time an AI answer could affect a person, decision or work product.