A student digs into a math problem that references his favorite superhero, the product of an AI program that has tailored the problem to his interests and ability.
A math teacher reviews an AI analysis of her teaching that highlights how often she prompted students to show their reasoning when solving a problem.
And a group of students work together to solve an equation while an AI-powered tool identifies and promotes each collaborative interaction among the group.
These scenarios are glimpses into what the AI-powered math class of the future might look like—one that new data show math teachers are not being prepared for. As AI rapidly advances and makes more inroads into classrooms, many math teachers are not receiving the professional development they need to be confident in their abilities to integrate AI into instruction, according to a February survey by the EdWeek Research Center.
While experts say that AI holds great promise for improving math instruction, teaching foundational math skills remains the top priority in schools across the country.
“We know that children learn math from being able to problem-solve, being able to use reasoning skills, critical thinking, having opportunities to collaborate with each other and talk about what they’re doing,” said Latrenda Knighten, the president of the National Council of Teachers of Mathematics and a former math teacher. “AI will not change any of those things.”
Math teachers at all levels must simultaneously grapple with big theoretical questions as well as the more practical, day-to-day ones: How do you keep pace with rapidly changing technologies? What are meaningful and age-appropriate ways to incorporate AI into the classroom? What kind of math do students even need to learn when free AI tools can solve complicated math problems for them? And how will AI change the future of math instruction?
The answers to those questions are still taking shape, but the future is less hazy than it was two years ago when a powerful and free version of ChatGPT was first made available for public use. At the time, AI was largely seen as a threat rather than an opportunity, with teachers paying the greatest attention to how to prevent students from using the technology to cheat. Now, teachers across the country are experimenting with a variety of AI tools, not just ChatGPT.
How AI can be used to improve teaching and learning
Even as education companies are rushing to embed artificial intelligence into their products, and more teachers are using free tools like ChatGPT, Copilot, and Gemini to assist with lesson preparation and administrative tasks, it can be hard for educators to imagine what a math classroom infused with artificial intelligence might look like.
Perhaps some of the most readily familiar examples for many educators are K-12-specific chatbot tools, like Khan Academy’s Khanmigo, which acts as a learning assistant or digital tutor. (Khan Academy receives funding from the Gates Foundation, which also provides support for Education Week’s coverage of math education. Education Week retains sole editorial control over its articles.)
Jennifer Cook, a 4th grade science and math teacher at First Avenue School in Newark, N.J., said she has come to rely on Khanmigo’s AI features during small group instruction. The program gives her recommendations on how to group students based on data it has analyzed on their strengths and weaknesses—a task that used to take as much as three hours of Cook’s time per week now takes three minutes.
The AI features also allow Cook to essentially clone herself, she said, so as she works with one group of students, the others can still get instructional help.
“It just gives them another tool to use to be a little bit more independent with doing some of their assignments and tasks,” she said. “I don’t have to keep stopping to answer all their questions. I can just focus on those six kids that I’m working with.”
The AI assistant feature is especially beneficial for students who can’t get homework help from family members, Cook said, either because they are too busy or they are unfamiliar with the content.
It’s worth noting that AI has many uses beyond teaching math content, said Jennifer Jacobs, an associate research professor in cognitive science at the University of Colorado, Boulder. She is part of a team developing and studying AI-powered technologies that promote student engagement within a math classroom and teacher reflection outside of it.
One AI tool the Colorado team is working on provides math teachers with AI-based feedback on their teaching, Jacobs said. The tool records a math teacher’s classroom lesson—analyzing and logging moments when the teacher used quality instructional practices, such as prompting students to provide their reasoning in their solving of a math problem. All those data are then displayed on a dashboard for teachers to explore, course-correct, and inform their approach moving forward, either on their own or in a coaching session, Jacobs said.
“We use AI to get conversations going with teachers and to have data about their classrooms that they can then reflect on, to talk about, use, and track changes over time in that data,” she added.
Another project the University of Colorado team is working on aims to strengthen students’ collaboration skills in their STEM classes. Called the Community Builder, or CoBi for short, the program records and analyzes students’ conversations in real time, listening for moments when students are, say, asking questions, inviting others to speak, and not dominating the conversation. Students eye a dashboard as they work, which displays a tree that grows a different colored flower as CoBi “notices” examples of collaboration. For example, a blue flower blooms as the program logs students being respectful to one another, and a red flower blooms as the CoBi catches examples of students moving their thinking forward.
This tool isn’t strictly teaching math, Jacobs pointed out, but it’s teaching important skills for being successful in math and the workforce.
“It is helping the kids to engage collaboratively with the math content,” she said. “It is looking at their discourse around the math content. It’s a really unique way of thinking about how to bring AI into the classroom.”
In the future, AI could power tools that allow teachers to create unique visualizations for math problems on the spot, said Sierra Noakes, the director of ed-tech evaluation for Digital Promise, a nonprofit that works on helping schools improve their use of technology. (Digital Promise receives funding from the Gates Foundation, which also provides support for Education Week’s coverage of math education. Education Week retains sole editorial control over its articles.)
“If you’re learning about surface area, ... I could see a world where a student might say, ‘What’s the perimeter of Charizard?’” a Pokemon game character, Noakes said. “I could not draw that. I would need AI’s help making the lines we would be adding up to summarize the perimeter.”
Those examples only scratch the surface of how AI might reshape math instruction in the near future. But even as the use of this technology continues to expand, a recent survey by the EdWeek Research Center shows that not all teachers are convinced it will be useful. And many say they are already out of their depth when it comes to harnessing AI.
Math teachers want AI training, but they aren’t getting it
Math teachers are skeptical about the benefits of incorporating AI into math instruction. Thirty-five percent reported that instructional tools that incorporate AI will have no impact on achievement in their school over the next 5 years, according to the EdWeek Research Center survey, and 1 in 5 believe AI-powered instructional tools will cause math achievement to decline.
That skepticism may be driven by the fact that the current AI chatbots in use often get things wrong, said Noakes.
But that doesn’t mean that math teachers don’t want to learn how to manage the technology. Nearly 7 in 10 educators who teach math said they have not yet received professional development on using artificial intelligence to teach that subject. Of those, 66 percent said in the survey that they would like training.
While the majority said their school or district permitted them to use AI to teach, half said their principal or supervisor has not encouraged them at all to experiment with existing AI tools to teach math.
It’s perhaps no surprise, then, that many teachers are not confident in their ability to integrate AI into their teaching. Four out of 10 math teachers in the survey rated their skills at this as “nonexistent.” Only about 1 in 10 rated their ability as good or excellent.
The first thing math teachers need, said Knighten from NCTM, is professional development, and not just from the company from which their district just purchased a new AI-tool. And they need time to just experiment, she said.
“I know that it’s hard to find additional time, but plan time to explore what’s out there,” she said. “We have to know what it can and can’t do. Give teachers time in their collaborative planning period to explore some of these tools, what are the limitations, what are the responses, so that they are not caught off guard.”
Same as their students, teachers need the skills to understand, evaluate, and use AI safely and ethically, including a basic understanding of how the technology works, said Noakes.
“I think that AI literacy is something any teacher regardless of subject area deserves to get a lot of PD around,” she said.
What math skills do students need in an AI-powered world?
Looking ahead five years, many math teachers predict that artificial intelligence will be integrated into middle and high school curricula, according to the EdWeek Research Center survey. Fifty-five percent believe that will be the case, while 41 percent predict AI will be integrated at all grade levels.
What, exactly, that will eventually look like is less clear. But both Knighten and Noakes say it should include AI literacy for students. Teachers in all subjects also have an important role to play in modeling good AI use and showing students that this is not an infallible, all-knowing technology, they said. AI’s outputs are often flawed or flat-out wrong.
Everyone agrees that advancements in AI—in particular the ability of free tools to solve complicated math problems—should not replace the development of foundational math skills.
Learning those skills will remain as important as ever, in large part, because they teach students a critical skill in the age of AI: problem-solving. That’s according to Po-Shen Loh, a math professor at Carnegie Mellon University, the founder of LIVE, a math-tutoring/instruction program, and the former coach of the U.S. International Math Olympiad team.
“Having that logical capacity, problem-solving capability, actually math is one of the best ways to build this,” he said. “If you go and do hard math problems that twist your brain into a knot, that actually gives you a way to learn how to explore lots and lots of chains of possibility.”
The most durable skill students can develop today is to be able to solve a problem they have never seen before, Loh said. To do that, he thinks teachers should focus more on solving the kinds of problems found in middle school math competitions, which require students to understand fractions, calculate percentages, and analyze some geometry.
“These are sources of quite creative math problems where you don’t need to know any trigonometry,” he said. “You don’t need to know what a logarithm is. You don’t need to know any calculus. But the questions, you’ll have to wrap your head around in all kinds of funny ways.”
Loh added that success in math relies not just on creative problem-solving abilities but also teamwork, communication, and cooperation, which are skills that will also serve students well in an AI-driven world.
“This human-to-human communication is the differentiator between you and some robot,” Loh said.
