Title: Beyond the Right Answer: How We Used AI to Make Math Thinking Visible
If you ask any math teacher what their biggest challenge is, you’ll rarely hear "getting students to find $x$." The real challenge is knowing how they got there. Did they truly grasp the underlying mathematical concept, or did they just execute a set of memorized steps?
In our secondary school math classes, we set out to answer a simple question: What if we could hear the thinking behind every student's answer?
Here is how our journey evolved from messy video uploads to using AI to transform math oracy and conceptual understanding in the classroom
The Early Days: Screen Recordings and Padlet Overload
To encourage students to articulate their reasoning, we initially turned to tools like Padlet. We asked students to record their screens, talk through their solutions, and give peer feedback.
While the engagement was great, we quickly hit a wall:
Technical Friction: Uploading large video files created severe lag and technical headaches.
Shaky Execution: Student recordings were often hard to follow or visually unsteady.
Superficial Feedback: Peer comments were often generic, and providing immediate, detailed teacher feedback to every single student was unsustainable.
We realized we needed a tool designed specifically to capture authentic thinking in real time without the administrative load.
Enter Snorkl: A Continuous Learning Loop
To bridge this gap, we introduced Snorkl—an EdTech tool built around AI-driven formative feedback. Instead of replacing human instruction, Snorkl created an active, iterative learning cycle:
Solve $\rightarrow$ Explain $\rightarrow$ Receive Feedback $\rightarrow$ Discuss $\rightarrow$ Reconstruct Understanding $\rightarrow$ Re-explain
Students use a stylus on an iPad or tablet to draw, highlight, and write out their solutions while talking through their thought processes. Snorkl’s AI listens to their voice, reads their work, and generates immediate, personalized feedback on both procedural accuracy and conceptual clarity.
How We Used It Across Math Topics:
Construct & Explain (A-Math): When differentiating trigonometric functions using the product rule, students wrote out steps while explaining their choices. The AI flagged missing steps or incomplete problem sets instantly.
Annotate & Explain (E-Math Geometry): While solving parallel line angle problems, students highlighted geometric diagrams while verbally stating their reasons (e.g., vertically opposite vs. alternate angles).
Draw & Justify (Graphs & Tangents): Students sketched tangents directly on function graphs to calculate gradients, receiving AI prompts to double-check their coordinate selections.
What We Learned (and Honest Challenges)
The student response was overwhelmingly positive. Many shared that "acting like a teacher" forced them to organize their thoughts, helping them understand why a method worked rather than just how to do it.
However, implementing AI in the math classroom comes with real-world considerations:
Hardware Constraints: Hardware matters—stylus sensitivity varies significantly across devices, affecting handwritten work.
Time Management: Running a full explanation and iteration loop takes time—often an entire period for a single complex problem.
AI Limitations: AI isn't perfect. It can occasionally misinterpret nuanced freehand drawings (like imprecise tangent lines) or penalize selective assignment setups.
The Takeaway for Educators
Using AI for assessment isn't about replacing the teacher; it's about building feedback literacy and metacognition. Snorkl acts as an initial sounding board, giving students immediate guidance so they can self-correct before presenting their reasoning to peers or the class.
By making student thinking visible, we no longer have to guess what happens between the question and the final answer—we can actually hear it
Have you experimented with voice oracy or AI feedback tools in your STEM classes? How do you tackle student math misconceptions? Let us know in the comments below!
No comments:
Post a Comment