What AI can't replace

The knowledge worker won by knowing more. AI changed the rules by making knowledge abundant.

For the past few decades, you could build a career & a thriving business by knowing something other people didn't.

A knowledge worker is someone who uses information as the main tool for doing the work they do. They often rely on formal education and are paid to solve complex problems.

Most solopreneurs I know are knowledge workers and face a problem: information is not as expensive as it used to be. It’s not as hard to access it as it used to be. So selling your course or coaching becomes harder. Selling your writing services becomes harder. Selling most things online is harder now.

It’s scary.

But there are three things you can do—three paths you can choose from—to transform your business and make it future-proof. AI won’t touch these for at least the next two decades, and I’ll prove it.

Whatever business you’re in now, you can set it up for success by choosing what to zoom in on out of these three.

But first, what’s a knowledge worker and why did we even need those guys?

The purpose of a knowledge worker was to boost productivity. As is with all workers, but the knowledge worker’s case was interesting because it outgrew its initial purpose.

Peter Drucker coined "knowledge worker" in The Landmarks of Tomorrow, at a moment when he was watching the previous productivity revolution — Taylorism and scientific management applied to manual labor — already winding down.

Scientific management (Taylorism) had already solved the manual-productivity problem in the early 1900s by breaking physical work into small, standardized, repeatable tasks that a supervisor could specify in advance and a worker could just execute — which is exactly how manufacturing got that "fifty-fold" productivity jump Drucker cites.

But that method only works when someone above the worker already knows what needs to be done. As organizations grew larger, more technical, and more specialized after WWII, the bottleneck stopped being "can we execute the task fast enough" and became "who figures out what the task should even be so we reach the goal”. And also “who decides what the goal should be.”

A knowledge worker's job, unlike a manual worker's, isn't handed down fully specified — they're paid to apply theory, judgment, and abstract knowledge to define the work, not just perform it.

Then the knowledge worker took over. By 1956, for the first time in American history, white-collar workers outnumbered blue-collar workers in the U.S. workforce. Drucker's later argument in Post-Capitalist Society (1993) pushes it one step further — that knowledge itself had become the primary economic resource, ahead of the classical trio of land, labor, and capital, which is a very different claim than "some jobs need more brains than others." It's a claim that the whole basis of wealth had shifted.

But it was always tied to productivity. Knowledge workers took over because they produced results that could not have been produced without them. That isn’t the case anymore.

How can you be more productive than a machine? Maybe you still can, sometimes, in some situations, but until when?

AI can produce or help produce results faster and often better than a human can, as long as the human works within well specified frameworks. And knowledge workers do have very well articulated frameworks.

People are pouring billions of dollars in AI to make it better and better, too. Anything can become better if you pour billions of dollars into figuring it out.

So what chance do we have? How can a knowledge worker—the expert with the education and the experience—be useful right now?

I sold a property a couple of months ago at double the price that the AI predicted. My sale made absolutely no sense to the model. It was way too expensive per square meter considering the old building and the “end of city” location.

But I’d noticed things the AI couldn’t notice. I noticed the emotion of people posting “I’m looking to buy…” posts in Facebook property groups. I noticed how jaded they sounded. How they couldn’t quite say what they were looking for, but they knew they haven’t found it.

And I’m a very specific node in the real estate market system. It’s my money and it’s my vision. Real estate agents sell what they have to sell. People who get rid of their properties rarely invest in renovating them. And my direct competition is often made of speculators who do low quality renovations and want to sell fast.

I’m not a designer. My husband isn’t a contractor. We’re not real estate agents. What we have, as background, is we’ve moved a lot. So we’ve lived in maybe 10+ different apartments, as tenants and as owners.

And we KNOW—and I can’t tell you enough how well we know—what people are looking for when they go to a Facebook group and write “I’ve looked at everything online; is there anything else you can offer me?”

Maybe the AI can figure it out after it asks a bunch of questions. Or maybe it never will because it doesn’t know what it’s like to enter a property and just feel off.

My property was expensive for its size and location, but it sold 2 months after I listed it, in a very difficult market (market is in downturn in Bulgaria right now).

So when I say innovation, I don’t mean necessarily coming up with the next great technology. Innovation is a scary word these days. How much more can you innovate, we live in a world of driverless cars and delivery robots and daily conversations with machines.

But what you can innovate on is offering people something they can’t even articulate. Reading between the lines. Using your experiences, the off feeling that a machine can never get. Reading a room is still something AI can’t do and it’ll be a while before it does (if ever).

It’s this illogical, chaotic innovation I’m talking about.

You need to be in the room to read the room. It’s the kind of research Aaron Sorkin says he uses when writing his movies. He says there’s research where you look for specific answers and then there’s research where you don’t know what you’re looking for. You’re just talking to people. You’re just in the room.

I don’t know where you room is, online or offline. But you know. Go there and read and talk to people. If you’re doing it or have done it, do it more. Do it until the way to innovate becomes obvious to you, even though the AI is telling you it won’t work.

You’ll get there, I promise!