Abstract: Information systems and computing educators have long inferred what students can do from what they hand in. Generative AI breaks that inference. When any student can produce a fluent schema, query, program, or report with almost no effort, the quality of a submission no longer separates those who understand from those who do not. We argue that this is a problem of assessment validity rather than academic integrity, and that prohibition, detection, and graduated-permission policies leave it unresolved. We propose three properties that make evidence of competence durable under AI assistance (substrate-tied, attention-tied, and interactively probeable) and six observational surfaces for gathering such evidence. We then consider what this asks of instructors, programs, and researchers, state what we are not claiming, and offer four testable expectations as an invitation to the community.

Keywords: Generative AI in education, Assessment, Competency, Academic integrity, Information systems education, Computing education

Download This Article: JISE2026v37n4pp448-454.pdf

Recommended Citation: Smith, T. C., & Zhang, X. (2026). Invited Editorial: If AI Can Produce the Work, What Should We Grade? AI-Durable Evidence of Competence in IS and Computing Education. Journal of Information Systems Education, 37(4), 448-454. https://doi.org/10.62273/GVGA1877