Abstract
Green and Giblin’s ‘systematic review’ of multiliteracies in the December 2025 issue of this journal concludes: there is ‘no evidence’ of impact. This paper replies with a three-layer analysis of that claim and the evidentiary regime behind it. Layer 1 assesses the review on its own terms, systematic review protocols and (quasi)experimental control/comparison designs, and finds fundamental methodological failures and a mismatch between multiliteracies and standardized literacy outcomes. Layer 2 locates the deeper problem in statistical survey psychometrics, arguing that tests, latent-trait inference and normalized scoring are reductive, costly and increasingly anachronistic, with consequential harms for learning and equity. Layer 3 advances an alternative paradigm, cyber-social learning research, enabled by integrated Contextual and Generative AI. Using the CyberScholar.ai environment as an illustration, we show how AI can analyse multimodal knowledge artefacts and learning-process traces at scale, deliver rich feedback that transposes quality into quantity and reframe assessment and research as transparent, continuous and longitudinal evidence trails.