Superficial ai learning and uneven appropriation among adult learners: Evidence from a virtual ai literacy route in Mexico

This study examines what adult learners’ files reveal about their engagement with generative artificial intelligence (AI) during a non-formal virtual training route in Chihuahua and Madera, Mexico. The route included five courses delivered between 2024 and 2025. Administrative records listed 80 unique participants, while 53 of them left observable work in the anonymized documentary inventory. The complete corpus comprised 445 files: 440 learner-produced items and five administrative reports consulted only to reconstruct the educational context. All learner files were inventoried by course and format, after which 20 contrasting cases were purposively selected for closer qualitative reading. An exploratory rubric considered five dimensions: task fulfillment, elaboration, the quality of prompting or AI interaction, visible human intervention, and applied transfer. In this article, superficial AI learning is not treated as a diagnosis or a stable learner trait. It describes a documentary pattern in which AI-supported activity is present, but the available files show little evidence of iteration, verification, contextualization, revision, or transfer. The archive contained markedly different traces of engagement. Some products recorded sustained development, prompt refinement, reorganization and use in a specific context; others reflected minimum compliance, sparse development, an opaque process or material that could not be interpreted reliably. These records do not reveal participants’ internal understanding or establish algorithmic dependence. They do, however, show how unevenly critical appropriation became visible in the course archive. The study contributes a documentary framework that brings AI literacy into conversation with information literacy, adult education and information development, while keeping the limits of retrospective educational records in view.