ACE-AF-08BeginnerTier 1 · Core
Understanding AI Limitations: Hallucinations, Bias and Errors
Recognise common AI mistakes, including made-up information, bias, and errors, before they affect your work.
What you’ll explore
- Explain what AI hallucinations are, why they occur, why confident wording is not evidence of accuracy, and recognise the four types: factual, citation, numerical and identity
- Sort AI tasks into low risk and high risk, and verify specific claims against the original source before they reach a client, a colleague or a document
- Describe how bias enters AI systems through training data, tell representation, measurement and aggregation bias apart, and test an output for a bias lean
- Identify structural, completeness, currency and extrapolation errors, and explain how errors differ from hallucinations
- Apply the five-question Output Audit to any AI response, choose a verification approach for the type of output, and build a personal AI verification checklist for their own role
Course outline
01Hallucinations: Why AI Makes Things Up and Sounds Sure While It Does60 min
02Hallucinations at Work: High-Risk Tasks, Low-Risk Tasks and How to Verify60 min
03Bias: How AI Reflects the Past, and How to Spot It60 min
04Errors: The Failures That Look Like Successes60 min
05The Output Audit: Evaluating AI Responses Before You Act on Them60 min
Who it’s for
Anyone who uses AI tools at work and relies on the outputs to inform decisions or communications Managers responsible for teams who use AI tools, because understanding failure modes is a leadership responsibility People in legal, compliance, finance and other high-stakes roles where AI errors carry real consequences