TOM DACCORD

View Original

Artificial Intelligence and Personalized Math Instruction

One of the most exciting benefits of AI lies in its ability to offer a tailored educational experience, adapting in real-time to the needs, strengths, and weaknesses of each student. Math teachers, for one, can harness AI capabilities to provide individualized math learning experiences that cater to the diverse needs of their students. These tools can adjust the pace of content delivery, the depth of the subject matter, and even the teaching style, based on real-time feedback from the student. High personalization can also help ensure that math lessons resonate more deeply with students, leading to better comprehension and heightened engagement.

Yet, the majority of Math teachers and their supervisors are reluctant to integrate AI into their instruction. In a survey conducted by the EdWeek Research Center from May 31 to June 9, nearly half of the educators who responded—43 percent—believed that students should be mandated to use paper and pencil in class to ensure they are doing their own work in math. Twenty-eight percent equated using AI to solve problems with plagiarism. And 14 percent were in favor of completely prohibiting the use of these tools.

But what if AI-driven personalized tutors could help students understand Math without divulging answers to math problems? What if these AI tools could prompt students to think and work through through every step of solving a math problem? What if, as result, students really enjoyed learning Math?

AI personalized tutors are tantalizingly close to all these abilities. AI Intelligent tutoring systems, for one, can mimic one-on-one math tutoring by providing guidance and feedback, and adjusting educational content in real time based on a student’s performance.

One such system is Khanmigo, developed by the nonprofit educational organization Khan Academy. (The Khan Academy was founded by Sal Khan, who began tutoring one of his cousins in mathematics and created math lessons in the form of short videos.) Khanmigo uses AI techniques like natural language understanding, speech recognition, and computer vision to interact with students via voice or text. It understands their questions and needs, provides them with relevant content and exercises, and gives them feedback and encouragement.

Khanmigo’s standout feature is its ability to guide students through the sequential steps or thought processes needed to solve a problem, rather than simply providing the answer. It encourages students to engage with the problem and reflect on their thinking. Another advantage of Khanmigo is that it can act as a teaching assistant for self-directed students, freeing up teachers to devote time to helping struggling students.

StepWiseMath is another intelligent tutoring system that offers a personalized learning experience for students studying mathematics. It utilizes AI to interact with students, comprehend their queries and requirements, supply them with relevant content and exercises, and provide feedback and encouragement. Students solve problems by submitting each step for evaluation and receiving immediate feedback. Students can ask for hints, and even click a button to see turn-by-turn directions at any point while trying to solve a problem. Upon finishing a problem, they receive a summary of their performance.

Sizzle is another AI-powered tool that aids students in solving math problems. Acting as a tutor chatbot, Sizzle provides a personalized approach to solving math equations by guiding students through each step of a problem. For instance, Sizzle features a "Try a Different Approach" option that allows users to probe alternative problem-solving methods. And if users find themselves stuck, they can ask Sizzle for a definition.

There are certainly limitations to AI personalized tutors. For one, they can make mistakes and might provide irrelevant information. But the introduction of GPT4 has brought forward “stunning results” on standardized math assessments, including SAT Math and AP Calculus. In addition, the OECD predicts that AI will soon completely master its test of adult numeracy and problem-solving skills. In other words, AI is quickly mastering K-12 math content.

And AI can deliver this content in ways that inspire math learning. For example, Duolingo, the world’s most popular learning app, has introduced Duolingo Math. It incorporates the core elements of the Duolingo pedagogical approach: short, interactive exercises, AI customized content, and an engaging yet challenging environment. In short, Duolingo Math presents bite-sized lessons in a gamified setting to make math learning enjoyable. (And if you’ve ever used Duolingo you know it can be laugh-out-loud fun.) As one user puts it: “This game is so fun and for kids of all ages. Kids will love math and will look forward to complete it.”

Furthermore, students (and parents) are increasingly open to being tutored by AI. A recent study revealed that a significant majority (85%) of students who had been tutored by both AI and humans found ChatGPT to be more effective. The study also found that 39% of high school and college students surveyed have completely replaced tutoring sessions with ChatGPT, and 30% of parents of students surveyed have done the same for their children.

Historically, math education has been dominated by decontextualized equations and repetitive worksheets. While foundational mathematical skills remain important, the advent of AI challenges the status quo. True mathematical prowess lies in the ability to discern patterns, conceptualize abstract ideas, and apply them innovatively to real-world scenarios. This nuanced math understanding is a challenge for machines, but a realm where humans thrive and excel.