Megan Mocko (University of Florida), Larry Lesser (The University of Texas at El Paso)
Abstract
We first describe two activities where students use AI to engage statistics process knowledge. One is a two-part activity (to be presented at ICOTS12) used in a flipped classroom, where students prompt an LLM for a pre-class investigation, then rate the helpfulness of the LLM's output. During class, after a mini-lecture on prompting styles, the students then work with a partner to create a new prompt, rate LLM output, and reflect on the process. The other activity (Mocko, Lesser, Lugo, & Shein 2026) has students create a statistics mnemonic for an exit ticket, with and without LLMs, then reflect on tradeoffs. These activities build on three critical components: reflection, prompt engineering, and a heuristic for critical ongoing evaluation. We’ll conclude with having participants adopt their own activity built on these characteristics. To optimize our two synchronous hours together, we will assume ability to write simple prompts for an LLM and completion of an advance reading (Walter, 2024).