In the rapidly evolving world of education technology, there’s a prevailing notion that most tools should be designed to reduce friction—making learning smooth, seamless, and uncomplicated. However, Prof. Rene Kizilcec from Cornell University’s Future of Learning Lab challenges this assumption. He argues that this reductionist approach may be misguided.
“Learning is not about feeling good,” Kizilcec asserts. “The emotion of learning is frustration. That’s the emotion that’s most predictive of learning.” This deceptively simple yet profound insight has critical consequences for the ways universities deploy AI tools, how educators design courses, and how the educational technology industry often conflates convenience with substantive learning.
Understanding the Role of Frustration in Learning
Founded by Kizilcec roughly seven and a half years ago, the Future of Learning Lab examines the intricate relationships between technology, education, and learning science, catering to learners across all age demographics—from primary school pupils to post-secondary students. The lab is engaged in projects that cover a diverse spectrum of applications, such as a national database of tutoring sessions, AI-driven clinical training tools for medical students, and a language-learning platform known as ChitterChatter.
The pivotal inquiry guiding these projects is: what constitutes effective teaching, and how can we integrate technology meaningfully? Kizilcec proposes an uncomfortable truth—that the discomfort interwoven in learning experiences, specifically frustration, can be transformative for educational outcomes.
The Pitfalls of “Low Friction” Learning Environments
This stance directly challenges many educational technologies designed with a “low friction” approach, such as Microsoft Copilot. Kizilcec raises concerns about these tools being misapplied in an educational context, as they often provide immediate answers without fostering critical thinking. “Copilot happily gives you the answers to all the assignments. It does not hold back,” Kizilcec emphasizes, critiquing how many universities, including Cornell, have adopted this tool without fully considering its implications for student learning.
Kizilcec clarifies that he is not dismissing the role of AI in education; rather, he advocates for a responsible integration of educational tools based on principles grounded in research and learning science. “What we should be giving to students is tools that are designed to support their learning,” he states. Emphasis should be placed on empowering educators to tailor these tools to specific course objectives.
Technological Innovations in Education
Among the AI solutions Kizilcec endorses is HiTA.ai, a cutting-edge platform that offers custom conversational support tailored to the needs of students and faculty. This platform is being actively used in courses such as INFO 4100: “Learning Analytics” and HADM 4205: “Real Estate Financial Modeling.” Unlike simplistic tools, HiTA facilitates learning by encouraging students to engage in productive struggle, offering guidance through hints rather than direct answers.
The National Tutoring Observatory Initiative
Further reflecting the lab’s philosophy, the National Tutoring Observatory aims to be the world’s most extensive repository of video and transcript data on tutoring interactions. Kizilcec highlights the pressing need for high-quality data to inform what effective teaching looks like. “If we don’t know what good teaching looks like, how can we train models to replicate that?” he questions.
Working alongside seven providers—including expert human tutors and AI voice systems—the lab is gathering a dataset referred to as Million Tutor Moves, which aims to document at least one million teacher-student interactions across various subjects and educational contexts. This endeavor is crucial for refining educational practices and enhancing machine learning models to benefit teaching and learning.
Applications in Medical Education and Language Learning
Doctoral students within the Future of Learning Lab are putting these principles into practice. One significant creation is MedSimAI, a platform that enables medical students to practice clinical conversations with AI-generated patients—an innovative method that allows students to refine their communication skills before facing real-life scenarios.
Another initiative, ChitterChatter, connects language learners with AI conversation partners, effectively reducing the anxiety commonly associated with speaking a new language. These projects exemplify how educational technology can address specific learner needs, fostering environments conducive to productive struggle.
The Enduring Importance of Human Teachers
Despite the growing influence of AI tools, Kizilcec remains steadfast in his belief in the irreplaceable role of human educators. “Teaching is really complex,” he affirms, noting that teachers engage with students on multiple emotional and developmental levels that systems cannot replicate. The challenge lies in leveraging technology to enhance the teaching experience while never underestimating the value of human connection and empathy in education.
Ultimately, Kizilcec advocates for technology that supplements traditional teaching, delivering personalized feedback that enhances learning without undermining the essential qualities of effective teaching. “It is useful to start from the core principles,” he states, “and then think about how these tools enhance them—and where they create risk.”
