Whether an AI's output is factually correct. Accuracy varies by topic familiarity, recency of information, and specificity of the claim. Always verify specific facts, numbers, dates, and citations in high-stakes work — fluent, confident AI output and fluent, confident wrong output look identical.
Accuracy
Lessons that cover this
- Knowing When to Trust the OutputA checking routine proportional to the stakes, so verification does not eat the time you saved.
- Measuring Whether It Actually WorkedUsage statistics are not evidence. Here is what to baseline, what to track, and what to say to a sceptic.
- AI for OperationsStatus reports, vendor comparisons, and the SOPs nobody has updated since 2023.
- Why AI Gets Things WrongHallucination is not a glitch to be patched. It is the same process that makes AI useful, running without a fact-check.
- Choosing Your First AI ProjectPick the boring one. Here is how to find it and how to know whether it worked.
- AI for FinanceExcellent at the narrative around the numbers. Not to be trusted with the numbers.
Related terms
This definition is part of a free, structured course on how AI actually works.
Start learning