From struggle to success: how the College of Education is using AI to enhance learning with productive failure

Seyedahmad Rahimi, Anthony Botelho, and Avery Closser develop AI-powered learning app with support from the Bill & Melinda Gates Foundation.

Students are encouraged to learn from their mistakes, but what if the lesson requires students to grapple with failure to find success? Seyedahmad Rahimi, Ph.D., along with co-PIs Anthony Botelho, Ph.D., and Avery Closser, Ph.D., is harnessing the power of failure by developing an AI-powered app with the help of a $451,563 grant from the Bill & Melinda Gates Foundation.

During his doctoral studies, Rahimi was introduced to a teaching method through the groundbreaking work of Manu Kapur, Ph.D., that intentionally utilizes failure to help students learn new concepts. Kapur dubbed the method “productive failure,” and Rahimi went on to design professional development workshops to train teachers in implementing this innovative teaching approach. However, workshop participants expressed concerns about the challenges of crafting problems that fit the productive failure model and effectively integrating the method into their classrooms.

With the rise of generative AI in 2023, Rahimi saw an opportunity to address these challenges using cutting-edge technology. Rahimi and Botelho worked with their students to develop an app that would make productive failure more accessible and practical for educators – the ProductiveMath app. The two assistant professors, in collaboration with Closser, applied for and secured a one-year grant from the Gates Foundation to advance the teacher-facing components of the project.

 The grant focuses on expanding the app as a free tool for teachers by assisting them with creating and modifying engaging math problems, preparing for possible student responses, facilitating discussions around learning math concepts, and more, using both large and small language learning models. The goal is to create an app for teachers and students to maximize learning through productive failure.

While productive failure sounds contradictory, it is an effective, evidence-based instructional approach. Kapur theorized that students would be more successful at comprehending math concepts if they were presented with problems intentionally designed for them to struggle and fail to solve before receiving direct instruction. The initial struggles prepare students cognitively to understand solutions and identify and learn from mistakes, strengthening their foundational knowledge. This method is being woven into curricula around the globe, and Kapur has agreed to work as a consultant on the ProductiveMath project.

“Productive failure has been shown to be quite effective, but very few teachers have used it because it’s difficult to generate math problems that are suitable for this approach,” Rahimi explains. “So we are using generative AI to create those problems and to help teachers facilitate the instructional discussions that are vital to the process.”

The app will use AI to generate narrative-based math problems that match students’ knowledge, the topic being taught, and their interests, making the problems more engaging and accessible. It will also support students’ creativity and teamwork while providing real-time help to students and teachers during and after the problem-solving process. The project aims to make productive failure an accessible learning strategy for teachers and students.

Rahimi, who was recently awarded the Diane E. Haines Teaching Excellence Award, also leads the Creative Learning through AI and Stealth Assessment in STEM (CLASS) Lab. He sees productive failure as a means to encourage creativity and collaboration in students as they work together in groups to analyze math problems from multiple angles in search of complex solutions. 

“Creativity as an outcome variable is essential to my research, and I try to bring creativity into all the learning environments I design,” Rahimi notes. “Collaboration, like creativity, is an essential skill students must develop and enhance. Productive failure fosters both. With further development and evaluation, ProductiveMath will provide real-time assessment and support for these critical skills.”

Looking towards the future, Rahimi hopes to grow the app beyond mathematics to incorporate other subjects and languages. He expresses immense gratitude for the support from the Gates Foundation in enhancing the teachers’ side of the app that will carry ProductiveMath through the next development phase.