Math, My Son, and the AI

 Date: May 14, 2025

 

Tonight was an experiment. I set out to teach my son some advanced number theory, a topic crucial for his upcoming IOQM exam – a beast of a competition, preliminary to the Olympiads. The catch? I don't know the solution to the specific problem I wanted him to tackle. My plan was to use AI, not as an answer key, but as a guide, a step-by-step problem-solving companion. I figured this way, he'd learn the process, the how to think, rather than just memorizing a solution.

The idea felt innovative. I envisioned the AI laying out a logical path, and I would be there, facilitating, prompting him to think through each step. A collaborative exploration of a difficult problem.

But the reality? It wasn't smooth. It felt... disjointed. The AI would present a piece of the solution, and I'd relay it, trying to turn it into a question for him. "Okay, the AI says consider this property... what do you think that means here?" My role felt less like a teacher and more like a reluctant translator, a medium he had to go through to interact with the 'real' source of knowledge, the AI.

There was a point where he was genuinely confused by an AI response. He looked at me, expecting the clear, expert explanation he's used to from his teachers. But I couldn't give it to him immediately. I had to understand his confusion, then figure out how to phrase it clearly enough for the AI to process, and then wait for its response. It was a clunky back-and-forth, a far cry from the fluid dialogue he'd have with a human expert who could instantly grasp his point of difficulty and re-explain in a different way. He's accustomed to that immediate feedback loop, and this wasn't it.

Honestly, there was an internal feeling of inadequacy, of not really doing anything meaningful. I wasn't using my own intellect or expertise in the way I'm used to when teaching. I was just... there, managing the interface. It's uncomfortable to feel like you're just a passive player in your own child's education, especially in a subject I care about. I want to be an active participant, guiding him with my own understanding, even if it's built with AI's help.

The experience solidified a thought: maybe the AI's primary role, for me, should be in my preparation. If I could use it beforehand to understand the problem and its solution deeply, then I could approach the teaching session with that knowledge. I could anticipate his questions, understand the AI's output instantly, and translate his doubts more effectively, or even answer them myself. That feels like a much more active, meaningful role.

The difficulty, of course, is finding the time and motivation to do that preparation before the teaching session. My original approach was partly born out of not having that prep time. But without it, we're both essentially students learning from the AI together. And frankly, given his focused preparation for the IOQM, he might even be faster at grasping things than I am in the moment, potentially slowing him down. That's not the goal.

Despite the awkwardness of the process tonight, I cannot deny the value the AI brought. It highlighted a specific property of a polynomial used in the problem that I wouldn't have easily identified. And when my son asked a question that was a slight extension of that property, the AI provided insights that were genuinely valuable, not just for this problem, but for others he might encounter in the future. Insights that might be hard to get from a standard textbook or even all human teachers.

We noted that property down, with the plan to revisit it later. I didn't delve deep into it during the session because I wanted to finish the original problem, and honestly, I wasn't sure how to apply that new property myself yet. I need a problem where it fits naturally to truly understand its utility. That's another task for the AI later – generating specific problems to practice these new insights. But that, too, needs to be done separately, validated before bringing it to him, adding another layer of time and effort outside the direct teaching moment.

So, the experiment was... illuminating. The AI is a powerful tool, capable of providing valuable, perhaps even unique, mathematical insights. But integrating it effectively into a personal teaching dynamic, where the human connection and the teacher's active role are preserved, is a complex challenge. It's not as simple as just putting the AI in the middle. It requires careful thought about how the AI is used, and perhaps, a shift in where the human effort is focused – maybe more on pre-session learning and less on live mediation.