SFAILQ, AICOS, MAIL-CS: Comparing Validated AI-Literacy Instruments
The field of AI-literacy measurement matured from "almost nothing" to "several validated scales" in two years. Anyone evaluating a program today no longer needs to build an instrument — they need to choose and use one correctly. That is the actual work.
1. The landscape (Guide 2, §2.1)
Our corpus documents at least nine psychometrically tested AI-literacy scales. The four most relevant for organizational programs:
| Instrument | Items / factors | Target group | Evidence base |
|---|---|---|---|
| MAIL-CS | 32 items, 4 factors (Foundational Knowledge & Ethics; Operational Skills; Critical Evaluation; Application & Innovation) | Chinese university students (N=850) | EFA/CFA split-sample; α=.91, ω=.92; convergent + discriminant validity |
| AICOS | Objective AI-competency scale, incl. GenAI dimension | Adults | Development + validation; objective format |
| SAIL4ALL | 56 items, 4 themes (What is AI? / What can AI do? / How does AI work? / How should AI be used?) | Adults, any setting | Psychometric validation; themes usable independently |
| SFAILQ | Six facets, adolescents to midlife adults | Multi-cohort | Development + validation (2026) |
2. Selection rules
- For organizational programs: prefer AICOS or MAIL-CS (objective/validated, GenAI-era).
- For cross-population benchmarking: SAIL4ALL (themes usable independently).
- For roles with domain context (educators, professionals): FALCON-AI (role-embedded).
- For behavior/reliance questions: MAIL-CS plus behavioral items.
3. The key insight: instrument choice changes the measured construct
MAIL-CS finds four factors; other scales find one dominant A-factor (44% variance, N=517). The instrument you choose determines what you call "AI literacy" — and therefore what your program counts as success. Choose before training, not after analysis.
4. Self-report is not enough
Objective and self-report measures correlate only weakly. Programs relying on pre/post self-assessment systematically overstate their impact. Pair an objective instrument with a task-based measure (see Self-report lies).
Further reading
- KI-Literacy Readiness Mini-Audit: free 10-minute self-check
- EU AI Act Art. 4: the legal obligation
- KI-Literacy-as-a-Service: from mini-audit to enterprise package
Sources: Guide 2 (Evaluation), §2.1 instruments; corpus of 16,459 papers (as of 2026-10). EOF echo "en1 done"