​Desirable Difficulties: Configuring Large Language Models for Effective Tutoring

Abstract

Large language models (LLMs) promise to boost productivity. But what about learning? Can learning be enhanced with these tools, or will that responsibility be offloaded? In this talk, I describe the importance of desirable difficulties and how critical "strategic friction" is throughout the learning process. In doing so, I outline key learning principles from a cognitive science perspective that can be applied as LLMs become integrated into educational contexts. To illustrate these points, I describe a recent (and still ongoing) independent learning project in which I have configured an LLM to act as a personal tutor that provides scaffolding, hints, and feedback without giving away the answer. While more formal research is needed, I will share my experience with this approach along with tips, tricks, and words of caution if you adopt a similar technique.

College

College of Liberal Arts

Department

Psychology

Location

Kryzsko, Solarium, Winona State University, Winona, Minnesota; United States

Start Date

4-23-2026 9:00 AM

End Date

4-23-2026 12:00 PM

Presentation Type

Event

Share

COinS
 
Apr 23rd, 9:00 AM Apr 23rd, 12:00 PM

​Desirable Difficulties: Configuring Large Language Models for Effective Tutoring

Kryzsko, Solarium, Winona State University, Winona, Minnesota; United States

Large language models (LLMs) promise to boost productivity. But what about learning? Can learning be enhanced with these tools, or will that responsibility be offloaded? In this talk, I describe the importance of desirable difficulties and how critical "strategic friction" is throughout the learning process. In doing so, I outline key learning principles from a cognitive science perspective that can be applied as LLMs become integrated into educational contexts. To illustrate these points, I describe a recent (and still ongoing) independent learning project in which I have configured an LLM to act as a personal tutor that provides scaffolding, hints, and feedback without giving away the answer. While more formal research is needed, I will share my experience with this approach along with tips, tricks, and words of caution if you adopt a similar technique.