My calculus lecturer taught me how to use AI twenty-five years ago. He said: "I want you to be lazy."

In 2001 I was sitting in a lecture theatre at UCT with two hundred other second year engineering students, taking MAM 280 — second year calculus. The course had a brutal reputation. Students failed it, repeated it, failed it again.

Our lecturer was Dr Reinhard. He didn't chase academic prestige or journal publications. He was, deeply and unmistakably, a teacher — enthusiastic about the subject, rigorous with his students. By the time I sat in front of him he had been teaching this course for fifteen years. He would go on for another decade, moulding the minds of perhaps ten thousand engineers.

He moved around the room, questioned students at random, tap tap tapped at the chalkboard with conviction. And regularly, to a room full of engineers who thought hard work meant grinding, he said:

"I want you to be lazy. Tell me how to do it in the laziest way possible."

He didn't mean cut corners. He meant the opposite.

Calculus is a language for describing reality. Its power is compression — boiling a system down to the simplest possible expression that holds the full truth. The best answer uses the fewest steps and holds the most truth. Maximum truth, minimum waste. That is elegance. That is what lazy means.

Twenty-five years later I think about this every time I open an AI tool.

The question is never "how do I work harder?" It's: what is the most elegant prompt — the fewest, most precise words — that holds the truth of what I actually need?

Sometimes the lazy answer is right there. AI has collapsed hours of admin into minutes, freed up my mental energy for what actually matters, made me sharper in how I communicate, and accelerated my engineering work dramatically. Most of this post was dictated while driving to collect a passport. Dead time, turned into something.

Sometimes the lazy answer isn't available yet. I tried to build an agent to scrape property listing sites for my home search — elegant idea, bot protection killed it. Some problems need multiple attempts. Some MAM 280 students had to take the course three times. The discipline isn't to abandon the lazy question because today's answer is "not yet." It's to keep asking it.

Here's the thing about calculators: when they arrived, they didn't make MAM 280 irrelevant. Calculators handle arithmetic. Real mathematics is understanding relationships, structure, the why underneath. Calculators took the grunt work. MAM 280 remained essential.

AI is the same. It will absorb the grunt work. The equivalent of MAM 280 — the high level thinking, the understanding of structure, the ability to ask the right question — that's not going anywhere.

Dr Reinhard taught us that the smartest thing you can do is find the elegant solution. Not the hardest path. The truest one.

Twenty five years later, that's still the lesson.