AI literacy, briefly

What is AI literacy?

Knowing enough about AI to use it well and to judge what it gives you. Here is the definition researchers use, what it covers, and why you can't measure it by asking people how good they are.

A definition

The researchers behind this test build on the definition by Duri Long and Brian Magerko (Markus et al., 2025). AI literacy is

“a set of competencies that enables individuals to critically evaluate AI technologies; communicate and collaborate effectively with AI; and use AI as a tool online, at home, and in the workplace” (Long & Magerko, 2020).

In plain words: you know roughly how AI works and where it goes wrong, you can work with it, and you can judge what it hands you.

It overlaps with digital, data and media literacy but isn't the same: AI literacy is specifically about knowing what AI systems can and can't do, and dealing with them critically and responsibly in practice (Markus et al., 2026).

It isn't only for people in tech. The test's authors call it a key competence well beyond technical jobs (Markus et al., 2026).

What it covers

The test splits AI literacy into six areas (Markus et al., 2025):

• Know & understand: the basic functions of AI and how AI applications are used.
• Use & apply: applying AI knowledge and tools in different situations.
• Detect: telling whether an application is based on AI or not.
• Ethics: fairness, responsibility, transparency and safety in the context of AI.
• Create: evaluating AI applications, predicting what they do and developing them.
• Generative AI: understanding, reflecting on and competently using AI that generates content.

Why asking people doesn't work

Most AI literacy measures so far have been self-ratings (Markus et al., 2025). But how good people think they are and what they know barely line up. In the norming study, scores on this test and self-rated AI literacy correlated at r = .05, close to no relationship at all (Markus et al., 2026).

That is why this test asks questions with right and wrong answers instead. It measures what you know about AI; how well you work with AI tools in practice is a different question (Markus et al., 2026).

Test what you know

Guess your result first, then answer 12 questions (or 51 in the long version) and see how close your guess was. Free, no signup.