From Skepticism to Mastery: Embracing Generative AI in Education

Lew Ludwig, Director for the Center for Learning and Teaching at Denison University

Addressing Common Concerns

As the director of our teaching center, I frequently encounter skepticism about generative AI during workshops at my college. Faculty members voice concerns about its inability to produce meaningful writing, its lack of contextual understanding and its frequent mathematical errors. If someone overheard these conversations but was unaware of the subject being generative AI, they may easily chalk this up as complaints about our students, the perennial “kids these days.”

Overcoming the Expert Blind Spot

Whenever I hear the “kids these days” discussion, I often note how long that colleague has been teaching. It seems the longer we’ve been at our craft, the less our incoming students know. So, are students getting weaker? Maybe, but I also argue that we, the faculty, are getting better.

Sprague and Stuart (2000) describe the stages of mastery development, starting with unconscious incompetence, moving to conscious incompetence and then to conscious competence. Mastery, or unconscious competence, takes years to achieve. As new teachers, we are usually at the conscious competence stage, needing focus to teach effectively. Over time, repeated teaching deepens our understanding until actions become second nature. However, this can lead to an "expert blind spot" (Nickerson, 1999), where we forget the challenges faced by novice learners - kids these days.

Armed with the knowledge of the expert blind spot, teachers can better assist novice learners by understanding

 their needs for developing mastery. Additionally, skilled educators should maintain the challenge level within students' Zone of Proximal Development (ZPD, Vygotsky), providing an optimal level of difficulty—not too easy to bore, nor too hard to overwhelm, but just within reach with the proper support. This balance ensures that students are adequately challenged and supported in their learning journey.

Let’s return to our opening observations. While many faculty members are fine poking holes in AI output to the point of ridicule, I believe most of us cringe when we watch characters like Professor Snape (Harry Potter) or Professor Kingsfield (The Paper Chase) do this to their suffering students. We would not treat our students in this way. When a student gives a wrong answer, we reflect on the response, try to understand where it came from, and then guide the student from this misunderstanding to understanding through carefully scaffolded questions that stay within the student’s ZPD.

"Use Our Inherent Teaching Skills To Guide It Toward The Correct Solution. By Doing So, We Can Better Leverage Ai's Potential To Enhance Learning And Teaching In Our Classrooms"

What would happen if we did the same when working with generative AI? Please keep in mind most of these models are trained to ‘make us happy.’ That is, they will do all they can to provide a response, even if it is made up, something referred to as ‘hallucinating.’ Just like we can guide our students to better understanding, we can guide generative AI to give better responses.

Practical Strategies for Using Generative AI

To understand what generative AI is capable of, educators should consider the following strategies:

Contextualize Prompts: Provide clear, detailed prompts that set the context for the AI. For example, instead of asking, "What does it mean for a place to be 'natural' or 'wild'?" provide context: "You are a student in an introductory environmental studies class. Explain how the concepts of 'natural' and 'wild' have evolved in the context of Western colonization."

Iterative Feedback: Treat interactions with AI as iterative. Just as we guide students from misunderstanding to understanding through scaffolded questions, we can refine AI responses by providing specific feedback and follow-up prompts.

Develop Prompting Skills: Recognize that we are novice users of AI and need to develop our prompting skills. There is no magic wand prompt that will give the exact response we want. Be patient and open to learning, just as we expect our students to be. This humility and willingness

 to improve will enhance the quality of AI-generated responses over time.

From Novice to Expert: Teaching AI with Patience

As educators, we must acknowledge that we are novices at prompting AI and improve our skills through practice. However, we are also skilled teachers who excel at guiding our students toward understanding. When working with AI, we should apply the same principles of patience and guidance. If an AI provides an incorrect or unsatisfactory response, don't dismiss it out of hand. Use our inherent teaching skills to guide it toward the correct solution. By doing so, we can better leverage AI's potential to enhance learning and teaching in our classrooms.