Research programme
AI literacy measurement
Whether AI literacy can be measured objectively, and whether the measure travels across languages.
- Studies
- 2
- Span
- 2026
- Published
- 0
- Under way
- 2
AI literacy is being defined and assessed faster than it is being validated. Most instruments in circulation are self-report, and self-report about a fast-moving technical domain has a known failure mode: the people least able to evaluate their own competence are the least able to report it accurately.
This programme tests that directly. One study compares what people say about their AI literacy against what an objective measure records, to establish whether the self-report instruments now in use are measuring competence or confidence. The second asks whether an objective measure survives translation, since a construct defined in English against English-language tools may not describe the same thing elsewhere.
Both studies are running rather than published. The reason for building the objective measure first is that a self-report scale can then be validated against it, whereas a set of self-report scales validated against each other establishes only that they agree. This is the newest line of work here, and the one where the measurement questions are least settled.
Studies in this programme
Do people know how AI literate they are?
Self-reported AI literacy against measured AI literacy, in the same people.
Does an objective AI literacy measure travel across languages?
Multi-group invariance of an objective AI literacy test across languages and countries.