Take a peek inside MATHia to see how we’re using AI to make students better at math.
Twenty-five years ago, a team of cognitive psychologists and computer scientists at Carnegie Mellon University partnered with math teachers at the Pittsburgh Public Schools to apply what was known about learning science to math education.
Dr. Steve Ritter was part of that team and became Carnegie Learning co-founder and Chief Scientist. He and his team of cognitive and computer scientists have been at it since 1998, compiling and acting on data to refine and revamp how students explore mathematics.
The result? Our flagship AI-driven math coaching solution, MATHia®. MATHia leverages symbolic and machine learning-based AI to continuously evaluate how students learn math and what support they need to succeed.
And MATHia isn’t like any other AI-based learning program. As Dr. Ritter says, “Most companies use AI to make computers smarter; we use AI to make students smarter.”
We’re proud to offer an AI solution that teachers can depend on to support their students and help with planning and instruction. Come on in and let us show you around MATHia!
What is MATHia?
MATHia is a one-to-one, online intelligent math coaching system for grades 6–12.
With MATHia, students get the support they need to uncover their inner math person: guided instruction and mastery exercises tailored to their individual student needs.
Teachers love MATHia for the easy-to-understand reports that actually give them insight into what should come next for each student and each class as a whole.
How does MATHia accomplish all of this? With loads of historical data and the help of AI that learns about how students learn.
AI that understands how students think
MATHia’s unique approach to AI uses machine learning to study how students interact with and think about math. Based on more than 25 years of data, MATHia informs us of the problem-solving strategies that students follow, where common mistakes are made, and how we can help students adjust their thinking to get it right next time—and every time after that.
An integral part of getting AI to understand the student learning experience is training MATHia to look at more than the student’s final answer. Instead, our trademarked AI solution analyzes the steps a student takes to reach that final answer and decides what support they need and don’t need based on their actions within the software.
After all, as math educators, we know that a student can show us an incorrect answer even though they’ve actually done most of the work correctly. Re-teaching skills a student has already mastered is a waste of time. That’s why we’ve trained MATHia’s AI to analyze discrete skills and focus its support where students actually need it.
MATHia’s AI individualizes feedback and support
An example of how MATHia uses AI to provide hyper-individualized support is this response to a student learning about multiplying fractions. Although the student knows quite a bit about fractions, they haven’t calculated the correct answer. Here’s how they thought about the problem:
As you can see, the student has incorrectly determined that ½ x ⅕ is 1. If you were to only see this on your student’s paper, without any insight into their thought process, how would you help them?
That’s a tough call! Does the student understand common denominators? Do they know the difference between multiplying and adding fractions? Where exactly did they go wrong? Re-teaching the whole process would be a waste of time and resources, so what now?
MATHia’s AI is here to help! Since students work through each problem step-by-step, it can determine where they need help, down to the granular skill level.
In this particular case, the student needs to review when to use the addition procedures and when to use the multiplication procedures.
Giving students conceptual knowledge of where these procedures come from helps them identify and carry out the correct procedure. As students work through problems in MATHia, they receive Just-in-Time Hints, a personalized number of problems, and instruction that solidifies conceptual understanding (like in our fractions example) and steers them on the correct path.
If another student struggles with a different step, say determining common denominators, their MATHia experience will reflect that specific need.
MATHia’s AI never provides arbitrary support—it’s always targeted and individualized so students don’t flounder in a Groundhog Day of learning the same skills over and over.
Another way MATHia individualizes student feedback is a nifty tool called the Skillometer. The Skillometer uses MATHia’s AI learning to help students track their progress through the discrete skills needed to solve a problem.
In this example, a student must master eight skills to show proficiency at identifying key characteristics of graphs of functions. The Skillometer circles fill as students demonstrate each skill, and MATHia’s AI serves up the correct amount and type of problems a student needs based on what they’ve done successfully and where they need more support.
Our AI adapts to how each student solves problems
As students interact with a MATHia Workspace (the smaller units that make up a course), our adaptive AI notes where and when students get tripped up and how they prefer to solve problems.
As students solve problems in the MATHia software, the AI notes their problem-solving preferences and is flexible enough to support them—though it may steer them to a more efficient method if one exists. It’s seriously like having a second teacher or math coach in the room, and what teacher can’t use that?
In this example, Emma and Aiden are in the same Geometry Workspace on transforming figures on the plane. Emma has started solving this problem by reflecting the shape, while Aiden prefers to translate the shape.
Emma and Aiden are using two different approaches to the same problem. MATHia’s AI can determine the strategy each student is using and the mistakes they made, and it adjusts the hints to each student's process.
How does MATHia’s AI help teachers?
Teachers' time is limited, and now, more than ever, they have more to accomplish every day. That’s why we’ve leveraged MATHia’s AI to develop reports that take some of the burden of time-consuming data collection off teachers’ shoulders. MATHia also uses AI to give next steps and planning support so teachers can quickly see what the data says needs to happen next.
LiveLab is one of our most popular MATHia features, and we think you’ll quickly see why.
Teachers encounter various situations when students get stuck as they use learning software. Some may seek assistance for minor issues, while others may err on the other side of the spectrum and not request help at all. It can be difficult to know what level of support a student needs, which is where LiveLab comes in.
LiveLab uses machine learning to distinguish when a student is making errors but progressing (this is the best zone of learning—productive struggle) from errors where the student is flailing and not likely to progress without intervention (unproductive struggle).
As students are using MATHia, the LiveLab dashboard signals teachers in real-time to focus on the unproductive strugglers since they can most benefit from the teacher’s help.
In Ms. Hernandez’s first period class, she can easily see that Sterling has been idle for five minutes and needs a nudge to get back on task.
Ms. Hernandez also sees that Madonna has a life preserver icon by her status—that means she’s experiencing unproductive struggle and needs help ASAP. She likely won’t master the current skills without the teacher stepping in.
Ms. Hernandez can click on Madonna’s life preserver, and MATHia’s AI will show her the exact remediation needed and other data that may impact her intervention with Madonna.
Let’s give a round of applause (because that’s how we pronounce APLSE!) for our next MATHia report. The Adaptive Personalized Learning Score (APLSE) Report is an AI-driven student report containing data you won’t find with any other online math coaching software. This report can be generated when MATHia is used with our Middle and High School Math Solutions.
The APLSE Report predicts how students will perform on end-of-course assessments. By analyzing how much time students have spent in MATHia, how many Workspaces they’ve mastered, the number of errors they’ve made, and the hints they’ve asked for, MATHia’s AI develops an APLSE score for each student.
If a student’s APLSE score is 70% or higher, we can safely predict they will pass end-of-course assessments. How do we know? We’ve conducted studies to confirm these claims in Florida, Ohio, California, Texas, and more.
In addition to the LiveLab and APLSE reports, MATHia’s AI develops student reports based on overall student progress within the course, activity during individual MATHia sessions, skill-by-skill within each Workspace, and grade-level content standards.
Combined, the immediate data and built-in recommendations across MATHia’s AI-developed reports give teachers a powerful, time-saving tool for deciding the next steps for each student and their class as a whole.
MATHia’s powerful AI-supported platform makes gathering, interpreting, and actually using student data to make instructional decisions easier than ever.
Let MATHia work for you
There’s no denying it—we’re in the age of AI hype (and overhype!). As we navigate the opportunities and risks AI brings, educators should only invest time and energy into AI tools that are proven to produce real impacts on student learning. MATHia has continually done just that. And with Dr. Ritter’s team at the helm of continuous improvements, MATHia is only going to get better—for students and educators.
Whether you’re looking for a complement to our Middle and High School Math Solutions or a supplemental solution to any 6–12 math curriculum, you can depend on MATHia to be more than just hype.
Let’s partner up!
Before joining Carnegie Learning's marketing team in 2022, Karen spent 16 years teaching mathematics and social studies in Ohio classrooms. She has a passion for inclusive education and believes that all learners can be meaningfully included in academic settings from day one. As a former math and special education teacher, she is excited to provide educators with the latest in best-practices content so that they can set all students on the path to becoming confident "math people."
Explore more related to this authorMost companies use AI to make computers smarter; we use AI to make students smarter.
Dr. Steve Ritter, Carnegie Learning co-founder and Chief Scientist
