What Is AI? A Practical Introduction to Tasks and Limits

Define the task before the technology

In everyday use, one system may help search, another may sort messages and a third may draft text. They are not the same service, and the word “intelligent” does not mean a system understands your situation as a person does. Ask what inputs it takes, what output it returns and how you could check it. UNESCO's AI competency framework separates understanding, application and creation; this guide starts with understanding and a limited practice.

What kinds of AI task might you meet?

A tool might classify data, estimate a value, recommend an option or generate content. These describe tasks, not a final catalogue of every AI technique. Summarising your own notes calls for different checks from analysing or generating an image. Do not choose a service merely because its name is popular. The AI tool selection guide helps compare purpose and limits without claiming that Practice is a generator.

What can be useful, and what might go wrong?

An AI tool may speed up a draft or organise repetitive information, but it may add an incorrect claim, miss important context or make it too easy to share data you should not upload. Speed is not accuracy. Pick a checkable task, keep responsibility for the decision clear and compare consequential claims with independent sources. This is an everyday application of evaluation, not a complete risk catalogue; the NIST AI Risk Management Framework explains why evaluation matters in responsible use.

Try one exercise to test your understanding

Write a non-sensitive task in one line: “Summarise these three general notes.” Define an output you can check: “Three sentences, each traceable to one note.” Try the clear-prompt practice, then compare the result with what you supplied. If the tool adds a claim, verify it independently or remove it. Use AI skills to improve the request and critical thinking to examine a claim.