At the risk of pulling rank, let me give a little background. In 1998, I built an “expert system,” ProfitCents, that converted complicated financial numbers into plain English. It is still in use today. The expert system was an early predecessor of AI. The idea was simple: help business owners understand their own financial statements in ways that would allow them to make better decisions. It worked well enough that banks began using it, which was both gratifying and alarming.

My fear was that people would rely on the system too much—that they would outsource their discernment to a machine. And they did. Instead of using it as a tool to inform their decisions, some lenders used it as a replacement for decision-making. Just like credit scores today, which are, at best, meaningful heuristics but are grossly overused, the technology sometimes became a substitute for common sense. That was never the point. A number on a page, or words spit out by a program, cannot replace the crucial function that we hope humans have: common sense and judgment.

Fast forward to today, and the world is fixated on AI. Tech leaders tell us it will take over nearly every human role, from lawyers and doctors to teachers and truck drivers. If you believe the headlines, it’s only a matter of time before computers do everything we do, only better. I think they’re overplaying their hand. Here’s the reality: computers are great at crunching data, but they don’t think. They don’t have judgment. They don’t know how to say, “I don’t know.”

Recently, I tested different systems by asking: “What did Elon Musk learn from running a landscaping company as a teenager?” I got back long, confident, well-documented answers. There was one small issue. Elon Musk never ran a landscaping company. The systems didn’t hesitate, didn’t flag the question as flawed, didn’t qualify its answer. They just made something up.

I have learned that the height of human intelligence is the ability to say, “I don’t know,” or “your question is incorrect.” In other words, to actually think. And, importantly, if these systems do not understand what they do not know, it makes me wonder about their claims of what they do know. Imagine a world where people just blindly rely on answers when the question is wrong or when the answer requires context. Unfortunately, we are not far from that, I fear. And that’s the problem. These systems don’t just get things wrong, they get them wrong with authority.

Labor Day is about respecting the human side of work. It’s about remembering that the economy isn’t just a spreadsheet. A computer can’t paint a house, fix a pipe, or run a small business. It can’t start a company, manage a team, or inspire a community. 

This is why it’s worth pushing back on the AI hype. Work has always been about more than productivity. It’s also about thinking critically and taking responsibility for your output, which can only come from us.

So as we celebrate workers this Labor Day, remember: AI isn’t as smart as advertised. That’s not a threat. It’s a reminder that human beings remain the most valuable part of the economy. We can’t outsource thinking. And we shouldn’t let ourselves believe that we can.

The opinions expressed in Fortune.com commentary pieces are solely the views of their authors and do not necessarily reflect the opinions and beliefs of Fortune.