The Math I Never Truly Learned

I sucked at math as a kid. Honestly, it was probably the subject I struggled with most my entire life. I spent years staring at blackboards feeling like I was not smart, and decided early that I wanted a career path that doesn't involve learning more advanced math.

When I thought about the possibility of AI replacing basic skill, how nobody needs to learn syntax, making prototypes, or do calculations anymore, it made me step back and wonder: why did we spend so much time learning math in the first place?

The standard defense teachers always gave us was that math isn't about crunching numbers, it's about learning how to think. It’s supposed to build logic, structured reasoning, and problem-solving muscle. But looking back, I honestly wonder how much of what we learned ever actually transferred to the rest of our lives.

Plenty of us ground our way through college calculus, linear algebra, or statistics. We learned to follow the steps, pass the exams, and check the degree requirement boxes. But the moment the syllabus ended, the practice stopped, and so did any real cognitive payoff. We walked away with some vague habits, like breaking a big problem into smaller pieces, but we never stayed in the game long enough to develop genuine mathematical taste. We never got that subconscious instinct a real mathematician has when an argument looks technically valid on paper, but just smells wrong. That kind of intuition lives years past the point where most of us checked out.

I'm afraid we are walking straight into the same trap with coding.

If the new benchmark for programming becomes "the AI generated this and I can roughly skim the file without getting lost," we are going to end up with a generation that has surface-level literacy, but zero real instinct.

You can’t prompt your way into knowing when an architecture is going to silently rot. And you definitely don't learn those lessons by reading generated solutions either. You only learn them by writing terrible code yourself, watching it collapse under its own weight in production, and pulling painful all-nighters cleaning up the mess.

The irony is that this specific, bruised kind of judgment is the exact thing you need to supervise an AI effectively. Without it, you aren't directing the tool, you're just nodding along to whatever it hands you.

I'm not saying that everyone needs to spend 40 hours a week writing boilerplate by hand. Maybe it’s just about keeping the deliberate reps alive. Doing fewer of them, but doing them on purpose, knowing that the exercise isn't about producing the file, but about keeping your own taste sharp.

After all, people still practice their scales, even in an age of synthesizers. I gave up on math too early, and I hope you stick with your craft a bit longer, whatever you're struggling to find meaning in right now.