Students’ perspectives of the use of Khanmigo as a mathematics AI tutor
Summary
This study explores college students’ experiences using an AI tutor to support their math learning. Specifically, we conducted a semi-structured interview of around 50 students in two sections of a college algebra class that had access to the assistance of Khanmigo, an AI-powered mathematics tutor by Khan Academy. The results show that students generally have a positive experience using Khanmigo and other generate AIs to complete their assignments and study for their exams. Overall, students have learned to appreciate the benefits of using this technology but also recognize and handle the limitations of these novel tools. Most of them are willing to continue using them to support their academic endeavors.
Drawing on students’ perspectives, we identified several benefits of using Khanmigo, such as being an immediate guide that can be accessed anytime, anywhere, as well as its ability to generate useful practice problems accompanied by clear, step-by-step explanations. In contrast, students also noted certain drawbacks, including the perception that responses were sometimes overly slow or excessively detailed, and that the AI struggled with more complex problems. Based on these insights, we offer recommendations for the effective integration of Khanmigo into classroom instruction. Overall, our findings provide a valuable foundation for reflecting on both, the promise and the challenges of generative AI as a supportive tool in students’ learning process.
Reflection
Generative AI is making a great impact on education. In my own tutoring practice, I observed that students are using AI tools like ChatGPT, Mathway or Photomath to solve their homework. While these tools can provide some help, I noticed some uninformed practices. For instance, students were frustrated they were not receiving correct answer from the tools unaware that they typed incorrect prompts. Meanwhile, others used the tools primarily as shortcuts to reach final solutions, missing the opportunity to engage in the problem-solving process. These habits limit their understanding and hinder the development of their mathematical reasoning. Recognizing this challenge, I wanted to explore the integration of AI into mathematical education and reflect on the tool’s benefits and demonstrate how to use it responsibly.
The study showed that generative AI has strong potential to serve as a mathematics. It offers a clear and useful explanation of mathematical problems, uses accessible language and simplifies complex ideas and can check the accuracy of students' answers. It also offers unique advantages that an in-person tutor cannot always provide, such as being available anytime and anywhere, offering endless patient and creating a stress-free way to ask questions without the pressure of speaking in person. These qualities can help students build confidence and gain mathematical independency. Hence, as educators and learners we should take advantage of these opportunities in the learning process.
At the same time, we must remain aware of AI’s limitations and potential drawbacks. For example, generative AI can create wrong or misleading answers and it can have a hard time with more complex problems. When this happens, students may be left with misconceptions or incomplete learning experiences. Moreover, using the AI to get answers instead of taking the time to solve the problem undermines true comprehension and mathematical independence, hence, the overreliance on AI makes the tool not a complement of the learning process but an obstacle. One way to address this challenge as educators is integrating tools like Khanmigo, which guide students step by step through problem-solving. This approach supports the learning process, and the comprehension is reinforced. On the learners’ side, we should develop a self awareness of our own learning, like recognizing when we need to push ourselves further solving a problem vs seeking the help of the AI to solve it.
This project offered me the opportunity to know first-hand how students interact with generative AI tools when learning mathematics and I hope that readers may also gain valuable insights into the role of AI technologies in mathematics education. For me, the experience reinforced the importance of balancing AI’s strengths with human guidance, while also ensuring that students learn to use these tools both effectively and responsibly. In the future, I plan to integrate one or two generative AI–based activities into my classroom with the purpose of supporting specific learning goals and foster independent problem-solving.
Artifacts
Below, you can check the preliminaries of the project and a summary on how to conduct a qualitative research through interviews.