Let the vibe be with you: A paradigm shift in statistics education?

Abstract The rapid evolution of artificial intelligence (AI), particularly large language models (LLMs), is reshaping how learners engage with computational tools. In software engineering, vibe-coding -a conversational paradigm for iteratively co-developing solutions with AI-has shown both productivity gains and risks of cognitive offloading. Its implications for statistics education, however, remain underexplored. We introduce vibe-modeling as a conceptual extension of this paradigm to statistical inquiry. Using the Julius AI agent as a case study, we examine how current AI tools can support an interactive, prompt-driven approach to model development. Through an illustrative analysis, we explore how such workflows may enable users to traverse alternative specifications-across likelihood families, effect scales, and conceptual framings-via iterative prompting. This dialogic process supports statistical sense-making by making modeling more exploratory, revisable, and visible than traditional linear workflows. At the same time, the case highlights important tensions: generative fluency can encourage model proliferation, obscure shifting assumptions, and create an illusion of methodological sophistication. We frame vibe-modeling as a promising direction that redistributes expertise toward contextual judgment, effect interpretation, and critical oversight of AI-generated outputs. While its overall pedagogical impact remains an open question, vibe-modeling may challenge and potentially reconfigure established practices in statistics education. When supported by appropriate instructional scaffolding, it may foster more reflective, inquiry-driven engagement with statistical modeling, particularly among novice learners.