Design considerations for teaching with AI-hallucination-inducing prompts
April 2026
University System of Maryland Generative AI Showcase, Virtual event
When using generative AI, it is important to know that LLMs occasionally hallucinate. For example, LLMs at times create and cite correctly formatted references to articles that do not actually exist. They can also produce well-written stories about recent events that contain fabricated details. At times, sentences within AI output can contradict each other. Hence, human expertise is still essential in judging the veracity and usefulness of AI outputs. One strategy that AI designers use to reduce hallucinations is to purposefully feed LLMs prompts that induce hallucinations and then subsequently refine the LLMs to avoid similar hallucinations in the future. While such hallucination-inducing prompts are helpful tools for AI platform designers, they are also valuable for teaching students about the affordances and constraints of AI. The purpose of this presentation is to delineate some design considerations for using AI-hallucination-inducing prompts in ways that promote thinking and reasoning among students across a variety of majors and disciplines.
Statistical knowledge for teaching in a data science era
June, 10 2024
2024 Electronic Conference on Teaching Statistics, Virtual Event
Statistical knowledge for teaching (SKT) frameworks provide guidance on what to include in statistics courses for prospective and practicing teachers. Existing SKT frameworks describe the content knowledge and pedagogical content knowledge needed to teach statistics. They also compare the knowledge needed to teach statistics to the knowledge needed to teach mathematics. Given current efforts to incorporate data science in Pre-K-12 curricula, it is important to extend SKT frameworks to consider how the knowledge needed to teach statistics compares to the knowledge needed to teach data science. This breakout session will offer conjectures about this matter. Participants will also be encouraged to offer their thoughts on how the knowledge needed to teach statistics compares to that needed to teach data science. The overall goal will be to identify essential experiences to include in statistics courses for teachers and update existing SKT frameworks for use in a data science era.
Contrasting cases in the development of statistical knowledge for teaching
April 2012
National Council of Teachers of Mathematics Research Presession, Philadelphia, PA
