Six-point check before you trust AI output

The six-point check means you must verify each of these before relying on AI output: accuracy of facts, sources cited or missing, hidden bias in the language, confidentiality of any data you fed in, tone matching your audience, and legal or institutional rules that might forbid AI use. A concrete contrast: treating AI text as a ready-to-submit draft is the mistake; treating it as a first draft you must edit and fact-check is the responsible step. The exam will test whether you know accountability stays with you, not the tool.

A useful trick: run each of the six checks as a quick mental filter. For accuracy, ask 'Can I verify this claim in two minutes?' If not, flag it. For bias, look for emotionally loaded words or unsupported generalisations. For confidentiality, never paste sensitive data into a public AI tool. An elimination trick: if a question offers 'just copy the AI output' as an option, it is always wrong. The correct answer will mention human oversight and verification.

To remember the six checks, use the mnemonic 'ABC TLC' — Accuracy, Bias, Confidentiality, then Tone, Legal, and Sources. Test yourself by imagining an AI-written email to a senior official: run each check in order. If any check fails, you must revise before use. This one compact test will tell you whether you understand the point: the AI is a helper, not the author.