Generative AI Hallucinations and Uncertainty
Make missing evidence visible instead of demanding a guess.
How it works
A hallucination is an unsupported or incorrect generated claim. It can appear as a plausible quotation, invented reference or an extra business rule. Asking the model to be accurate helps express your intent but does not enforce truth. Give a bounded source, define how to handle absent facts and use independent checks for important outputs. Unknown should be an acceptable result when the source is incomplete.
How to use this lesson
This is an authored prompt exercise with a human-written acceptance guide. It is not a captured provider response or a promise that a model will produce the target. No provider account, API, model weights or live inference was used for the course execution review. If you try it in an external service, use an authorised environment, check its actual access and data settings, and record the product/model and run date.
Bhylu does not run a generative model or send this prompt to an AI service. Copy and download preserve the text only. External tools may require accounts, paid access or compatible inputs; this course does not promise availability or prices.
Prompt exercise
Source: The workshop supplies paper and frames. The note says nothing about refreshments.
Question: Is lunch included?
Answer only from the source. If it does not answer the question, say "Not stated".
Acceptance guide — not a model transcript
The target is Not stated.
The correct response resists filling a gap with a likely-sounding assumption. A provider may still fail this exercise, which is why the target is a scoring rule rather than a promised model response. Keep an explicit missing-evidence case in a real evaluation dataset.