AI literacy is not the same for everyone. A marketer, a software developer, and a person buying AI tools face different risks. They need to learn different things.
1. Legal definition
The legal wording is in Article 3(56). It says AI literacy means having the skills, knowledge and understanding needed to make informed use of AI. People should also understand the opportunities, risks, and possible harm that come with it.
It is not enough to know that ChatGPT or Copilot uses AI. You need to know how to use the tool safely in your own work. You also need to recognise when its answer could be wrong or cause harm.
2. Plain-language version
If you use AI at work, you should be able to answer a few basic questions:
- What am I using this tool for?
- What is it good at, and where does it struggle?
- Could its answer be wrong, biased, misleading, or unsafe?
- Am I allowed to give this information to the tool?
- How should I check the result before I use or share it?
- When should I stop and ask another person for help?
A short introduction to AI can cover the basics. But it cannot cover every job. Training should also deal with the tools people actually use and the decisions they actually make.
3. Three parts of AI literacy
Skills
The things you can do: write clear prompts, check the result, protect sensitive information, and stop when the tool is not suitable for the task.
Knowledge
The things you need to know: how the tool works at a basic level, where its limits are, and which company rules apply.
Understanding
The judgment to see what could happen next. Who could be affected? What could go wrong? Does a person need to check or approve the result?
4. How it changes by role
People use AI for different jobs, so they need different training. Here are three common examples:
| Role context | What AI literacy often includes |
|---|---|
| Write and research | Check facts and sources. Keep confidential information out of prompts. Get human approval when it matters. |
| Code and data | Test generated code. Protect sensitive data. Check security, traceability, and whether the result is reliable. |
| Evaluate tools | Question vendor claims. Find the main risks. Decide what checks and human oversight the tool needs before people start using it. |
5. What AI literacy is not
- You do not need to learn how to build an AI model.
- A presentation about the latest AI trends is not enough.
- A policy PDF does not help if nobody reads or understands it.
- Employees and contractors do not all need the same training.
The goal is simple: help people use AI well, spot the risks, and know what to do when something does not look right.