What an AI assistant can and cannot do
Learn
A conversational AI tool can outline ideas, transform writing, explain concepts, compare alternatives, summarise supplied text and help build first drafts. Its usefulness improves when you give relevant context. It may also make up a source, misunderstand a request, miss an exception or reproduce an unstated assumption. These errors are sometimes called hallucinations, but a simpler rule works for beginners: never mistake confidence for proof.
Different tools may support web browsing, file analysis, images or connected services. Do not assume that every tool, account or session has the same access. If a task depends on a current regulation, local price, technical specification or an internal document, check that the assistant can consult the required source. Otherwise obtain the information separately.
Worked example
Prompt: "Tell me the current price of this tool." Without the exact product, region, currency and a current price source, the answer could be misleading. A stronger task would supply a verified quotation and ask: "Compare these two quotations using only the attached figures; flag missing terms."
Practice
Make a two-column table titled AI helps and Human must check. Include six tasks, such as drafting an email, confirming a legal requirement, generating quiz questions, calculating a budget and assessing a safety risk. Discuss what evidence would be needed before relying on the result.
Key takeaway
Match the level of verification to the consequences of being wrong.
Which information is safest for a first practice prompt?
Progress in this free study edition is saved in this browser only. It does not constitute a final exam result or issue a certificate.