MathBot: Transforming Online Resources for Learning Math into Conversational Interactions

Online math education often lacks key features of in-person instruction, such as personalized feedback. To emulate such interactivity while preserving the scalability of online systems, we developed MathBot, an automated text-based tutor that explains math concepts, provides practice questions, and offers students tailored feedback. We evaluated MathBot through two online studies conducted with Amazon Mechanical Turk in which participants learned about arithmetic sequences. In the first study, we examined user preferences, comparing MathBot with videos and written tutorials from Khan Academy. This within-subject study revealed that 42% of individuals in our sample preferred MathBot over videos, while 47% preferred MathBot over written tutorials. In a second, between-subject randomized study, we found that both MathBot and Khan Academy produced sizable learning gains, with MathBot performing slightly better, though the difference was not statistically significant. Our findings indicate that conversational agents are a promising tool for complementing online math education.