Rethinking Assessment in the Age of Generative AI: Process-Evidence Assessment for Information Systems Courses

Generative artificial intelligence (GenAI) has weakened a long-standing assumption in Information Systems (IS) assessment: that a competent submitted artifact reliably demonstrates student understanding. When students can use large language models to generate code, data models, requirements analyses, and written explanations, artifact quality alone becomes insufficient evidence of learning. This paper introduces Process-Evidence Assessment (PEA), a pedagogical framework that reorients IS assessment toward observable evidence of reasoning, verification, iteration, and explanation. PEA integrates assessment validity, dual-process theory, authentic assessment, evaluative judgment, self-regulated learning, and computing education into five components: assignment-level AI-use stratification, structured evidence packs, process timelines, reasoning-weighted rubrics, and written mini-defense questions. The paper reports formative classroom observations from a partial PEA implementation across three undergraduate IS courses in Spring 2026. The observations suggest that PEA made differences in student reasoning more visible than product-only grading would have, especially when students verified AI outputs, documented rejected suggestions, explained design tradeoffs, and answered submission-specific follow-up questions.