Let Humans Do Human Work

On Redwoods, Category Errors, and the Work That Actually Matters

Eric Markowitz wrote a beautiful essay this week called “It Was Never About AI.” You should read it. He walks through the redwoods with a friend, meditates on the feedback loop between Wall Street and Silicon Valley, and arrives at a line that stopped me cold:

We are not our tools. We never have been.

He is right. And he is also, I think, missing half the picture.

Markowitz describes a world where a 26-year-old quant analyst writes a note, a stock drops, 3,000 people get a calendar invite from HR titled “Quick Chat.” He describes the founder in the fleece vest preaching about empowering humanity while building products designed to make humans unnecessary. He describes the religion of optimization.

I recognize that world. I live adjacent to it. I run an “AI” company. And I want to offer a different frame.

The problem is not that we have powerful tools. The problem is that we have confused which work belongs to humans and which work belongs to machines.

I spend my days inside wealth management firms. These are businesses built entirely on trust. A financial advisor’s job, at its core, is to sit across from another human being and help them make decisions about their life. About retirement. About their children’s education. About what happens to their money when they die. This is human work. It requires judgment, empathy, experience, and the kind of pattern recognition that only comes from having lived through a few market cycles and a few difficult conversations.

But here is what else happens inside those firms. Someone spends four hours reformatting a report. Someone else manually enters the same data into three different systems. An operations associate burns an entire afternoon chasing a custodian for a document that should have arrived automatically. A compliance officer reviews a hundred emails by hand for a regulatory audit.

This is not human work. This is machine work being done by humans. And it is destroying them.

Not in the dramatic, dystopian way that makes for good Substack essays. In the quiet, grinding, soul-draining way that turns talented people into button-pushers. In the way that makes a skilled advisor spend 60% of their week on tasks that have nothing to do with the reason they got into this profession. In the way that creates burnout not from thinking too hard, but from not being allowed to think at all.

Markowitz writes about a founder who looks at AI and says: “This is a tool, and I will decide how it serves us.” I agree with every word of that sentence. But I want to push on what “serves us” actually means.

Serving us does not mean keeping humans in jobs that were never meant for humans.

There is a strange nostalgia embedded in the anti-AI argument. A sense that any job, simply because a person currently does it, is therefore meaningful human work. That preserving the job is the same as preserving the dignity. I understand the impulse. But I think it gets the causality backwards.

The dignity is not in the task. The dignity is in the judgment.

A financial advisor who spends her day exercising judgment, building relationships, navigating complexity, earning trust. That is dignified work. The same advisor spending her evening copying data between spreadsheets because nobody built her a better system. That is not dignity. That is a failure of imagination.

There is an old idea in philosophy called the “category error.” It means confusing one kind of thing for another. Mistaking a description for an explanation. Treating a metaphor as a literal truth.

I think we are making a category error about work.

We have lumped all labor into one pile and called it “jobs.” Then when someone proposes automating part of that pile, we react as though they are proposing the elimination of human purpose itself. But purpose and process are not the same thing. The work that gives us meaning and the work that merely fills our hours often look nothing alike.

The question is not: should machines do work?

The question is: what work should only humans do? And once we answer that clearly, how do we free humans to do more of it?

Markowitz invokes the redwoods. I want to stay with that image for a moment, because I think he is onto something deeper than he realizes.

A redwood forest is not efficient. It is redundant, overlapping, slow. By every metric a management consultant would use, it is poorly optimized. And yet it has survived for millennia.

Why? Because it allocates resources according to their nature. Roots do root work. Bark does bark work. Mycorrhizal networks move nutrients where they are needed. Nothing in the forest is doing another organism’s job. The system works because each component does what it was designed to do.

Now look at the average enterprise. Highly educated professionals doing data entry. Creative minds trapped in compliance checklists. Relationship builders buried in operational overhead. This is not an ecosystem. It is a misallocation. And the answer is not to cut the humans. The answer is to stop wasting them.

I should be transparent about my bias. I build technology that removes operational work inside financial firms. I have skin in this game.

But I did not start this company because I wanted fewer humans in wealth management. I started it because I kept meeting brilliant advisors who were drowning in work that had nothing to do with their clients. They did not need fewer people. They needed their people freed up to do the work that actually matters.

The firms I admire most are not cutting headcount. They are redeploying capacity. They are taking the countless hours a week their teams spend on machine work and redirecting that time toward clients, toward strategy, toward the kind of deep thinking that no AI can replicate.

They are not replacing humans with machines. They are replacing machine work with machines, and giving humans back their humanity.

Markowitz ends his essay with a declaration: we are not our tools.

I want to add a corollary: we are not our tasks, either.

Your job title is not your identity. The processes you execute are not your purpose. The spreadsheet you maintain is not your legacy. You are the judgment you bring. The relationships you build. The trust you earn. The decisions you make when the data is ambiguous and the stakes are real and there is no prompt you can write that will tell you what to do.

That is human work. Everything else is overhead.

And for the first time in history, we have the technology to make that distinction real. Not to eliminate humans. To liberate them.

The only question is whether we are wise enough to use it that way.

I think we are. But only if we stop arguing about whether machines should do work, and start asking much harder questions about which work deserves a human in the first place.

The redwoods are patient. They will wait for us to figure it out.