Abstract: Software development is shifting from line-by-line authorship to intent-mediated workflows (“vibe coding”) that require developers to orchestrate, evaluate, and integrate model outputs. This paper argues that information systems curricula should pivot from producing more code to reasoning more effectively in AI-assisted coding. This theory-informed descriptive study synthesizes trust-in-automation research with active-learning frameworks to reframe AI-assisted coding as a calibration problem of aligning reliance with demonstrated system capability, rather than a generation problem. Descriptive evidence from the 2023–2025 Stack Overflow Developer Survey provides motivational context: adoption is rising while stated trust declines (a widening usage-trust gap); trust differences across experience levels narrow by 2025; frustrations concentrate in accuracy, explanation, and integration; learning occurs primarily through informal channels; and vibe coding is broadly shared across experience levels. Integrating these patterns with theory, four low-overhead activity types are proposed for pilot testing in programming-intensive IS courses: (1) calibration sprints that move “almost-right” outputs to verified solutions, (2) spec-to-code exercises with explicit constraints and rationales, (3) integration diaries with lightweight automated checks, and (4) structured code reviews with roles and explicit accept/revise/reject decisions. The goal is to equip students to use fluent but fallible systems with calibrated reliance and reproducible process, keeping coursework aligned with evolving development practice.

Keywords: Natural language programming, Generative AI, Information systems education, Human-computer interaction (HCI)

Download This Article: JISE2026v37n4pp507-525.pdf

Recommended Citation: Bohr, J. (2026). From Line-By-Line to Vibe Coding: Teaching Orchestration in Information Systems Education. Journal of Information Systems Education, 37(4), 507-525. https://doi.org/10.62273/GFKF5366