AI and a New Era of Math Tutoring: Have We Seen This Movie Before?
Dr. John Woodward is a nationally recognized mathematics author, writer, and speaker. He is the past dean of the school of education and professor emeritus at the University of Puget Sound in Tacoma, WA.
As a researcher, he focused on mathematics interventions for academically low-achieving students, particularly in elementary and middle grades. Dr. Woodward has published more than 80 articles and presented on mathematics education issues throughout the U.S., as well as in Canada, Asia, and Europe. He is the senior author of TransMath, a math intervention program for middle school students. He also is the co-developer of NUMBERS, a math professional development program for K–8 teachers.
You’d have to been living beneath a rock for the past year to have avoided the onslaught of news about the future of artificial intelligence. AI will create unheralded prosperity. It will take most people’s jobs. It poses a serious threat to our existence. And by the way, it’s about to usher in an era of online tutoring that approaches, if not exceeds, human capabilities. Some like Sam Altman, CEO of OpenAI and ChatGPT, claim the latter—free online tutoring—will occur by 2030.
If you’re confused, welcome to the club. My career in education began with the appearance of the microcomputer. Beyond all of the initial hype, you could even see at that time that in respect to mathematics education, microcomputers were more than just an extension of calculators. Early spreadsheet programs in the 1980s like VisiCalc and Lotus 1-2-3 demonstrated that. Those programs were the tip of what was to come, that increasingly computers would largely replace the highly procedural calculations American educators were spending so much time teaching. Even into the middle of that decade, many fourth grade students were spending most of their daily half hour of math learning long division. That was unsustainable even in light of these nascent technologies.
Predicting the future is a dubious exercise. But looking backwards throughout the past 45 years offers thematic insights into what may come of AI and math education. Ironically, one of my first research grants from the U.S. Department of Education in the late 1980s was to create an AI-based system for diagnosing student misconceptions in math. The experience was instructive in terms of a central theme from 1980 to the early 2000s—educational computing was overhyped and underpowered. But more importantly, using a computer to understand what students might be thinking was a formidable challenge.
In the two decades before 2000, educators were preoccupied with the microcomputer itself and what eventually amounted to drill-and-practice programs as well as teaching productivity tools such as word processing. Since 2000, “the impact of technology on education” has become more diffuse. Attempts to show how technology can improve math performance, for example, are now largely peripheral to (or subsumed by) the Internet, smartboards, digital projectors, and of course, smart phones. Much of the talk today about AI and human tutoring returns us to that first preoccupation: Computers as tools to improve math performance, albeit at a much higher level than what drill-and-practice programs could ever achieve.
Arguably, some of the seeds of today’s AI discussions can be found in the mid-2010s. Two themes stand out. First, once powerful computing devices were networked to an Internet full of data, new possibilities emerged: machine learning, image and voice recognition, robotics, and more. In education, the power of distributed computing appeared in Google’s support for the Khan Academy, which Google’s CEO at the time described in revolutionary terms not unlike what you hear about the potential of AI for math tutoring today. Again, overhyped?
A second, and seemingly contradictory theme also emerged from critics who noted these advanced technologies may not always improve our lives, but instead, dumb them down and possibly make life, at times, more dangerous. By 2023, the “dumbing down” theme erupted in every high school English class across the country through ChatGPT. It could also be found with students struggling with their math homework, only to take pictures of the problems and get the answers effortlessly.
All of this brings us to now and the near term. As a guest on the EDVIEW360 podcast, our conversation examines the predicted world of AI and math tutoring through the lens of recent history. It focuses on terms like “tutoring” and “curriculum,” which are often used in ill-defined ways. The highly promising results from research in AI tutoring in math conducted as recently as last summer need to be examined in light of a rich history of curriculum design, learning, and motivational theory. Put simply, it is one thing to succeed with students who are struggling with what was taught this week in a math class. It is an entirely different matter to tutor—or more precisely, remediate—students who are months or years behind grade level in the subject.
Finally, if you want a preview before listening to the podcast, just ask ChatGPT, “What is Sam Altman’s vision of math tutoring and AI by 2030?”
