Live AR-Generated Tutorials Beat Pre-Authored Guidance on Task Quality in a 24-Person Study

arXiv·low signalXBlueskyLinkedInCopy link

"Generative Tutorial" (arXiv 2609.24955, 21 Sep 2026) targets the mismatch where visual instructions are authored in one context and followed in another, forcing users to translate demonstrated tools, materials and spatial relationships into their own environment. A formative evaluation of current image and video generation across 15 physical tasks catalogs the failure modes, and the resulting AR prototype proactively generates goal images and demonstration videos from observed workspace context and the predicted visual outcome of preceding actions. A 24-participant lab study found higher task performance quality, greater perceived workspace correspondence and shorter step-confirmation intervals than pre-authored guidance, with qualitative findings on how contextual resemblance shapes trust and generation errors distort interpretation.