Podcast Episode: Are We Preparing Students for What’s Coming?

Here’s a question that doesn’t get asked enough: if the software is easier to build than ever, what exactly are we still teaching people to do?

That’s the thread running through my recent writing on Unconventional Observations — who gets to innovate now, what expertise actually means in an AI-assisted world, and whether our institutions are keeping pace.

“For decades, organizations hired people who understood technology and taught them about the business. We’re entering a period when the reverse is becoming increasingly practical: people who understand the business can use AI to help build the technology.”

That’s not a small inversion. It means a finance graduate could build a forecasting tool, a marketing graduate could develop customer-behavior analysis — without first hiring a development team. The barrier that once required capital and technical headcount is shrinking.

And the examples aren’t hypothetical. There’s work happening in mining, healthcare, and sales — AI-assisted systems built with a roughly 700-page development playbook and 22 specialized agents handling architecture, testing, and deployment. The subject-matter expert still determines whether the output is actually useful. The technology lowers the floor; it doesn’t replace the judgment.

Which is where the post gets uncomfortable. If AI handles the entry-level work — the financial statements a young accountant grinds through, the research a new consultant does by hand — those people may never build the judgment that makes them good at their jobs later.

So we get productivity gains and a potential expertise vacuum at the same time. That’s a cheerful combination.

The post puts it plainly: “Producing an answer isn’t the same as understanding whether it’s correct. And understanding whether it’s correct isn’t always the same as knowing what to do with it.” Universities and employers both get called out for not taking that seriously enough.