In my entrepreneurship classes, students ask what will be left for them as machines become capable of the work they’re preparing to do. It’s a reasonable question, and a hard one to answer honestly. They need to make a living. They want the effort they’re putting into their education to lead somewhere, and underneath the career question sits something more personal: what happens to the life I’m trying to build if the things I get good at stop being scarce?
The usual answer is to keep moving. Learn the tools, develop the next skill, find something machines still struggle with and get good at that. Much of this is sensible advice. But we’ve allowed advice about earning a living to carry a weight it can’t bear. We’re asking it to explain why a person matters.
That’s the question behind my new book, What the Machine Cannot Do: The Irreplaceable Value of Being Human in an Age of AI.
A Reassurance That Can Survive Improvement
The title might suggest a list of things AI will never master: creativity, perhaps, or empathy, or the kind of judgment that only comes from experience. Those capabilities matter, and there are serious questions about how well particular systems perform them. But a defense of human worth built on the latest limitation of a machine has an expiration date attached to it.
The next release arrives, and a system does something we thought was safely ours. We retreat to another ability and tell ourselves this one will hold. Meanwhile, a young person is left wondering how many retreats there’ll be before we run out of reassuring things to say.
I want an answer that still works if the machines get much better than we expect. If a system writes a better report than I do, the report may be better. That tells us something useful about the report. It doesn’t establish that my children have less reason to love me, or that I have fewer obligations to the people who depend on me.
Losing work can be devastating. A declaration of dignity won’t cover the mortgage, and leaders can’t use philosophy to dodge the practical consequences of their decisions. But a person’s economic circumstances and the value of that person’s life can’t be allowed to become the same question. We have to take both seriously.
The Person Behind the Result
The book opens with something my father did. A junior employee at his company lost a child to a terminal illness and couldn’t afford the funeral. My father quietly paid for it.
A machine could help administer that kind of assistance. It might even catch a need that a busy employer had missed, and I’d welcome that. The burden relieved would be real.
But what stayed with me was that my father chose to carry some of it himself. Another person’s suffering changed what he did with his resources, and knowing that about him has kept asking something of me ever since. I can admire his generosity indefinitely without becoming generous myself. At some point, someone else’s need has to change what I do.
That’s one meaning of the word “cannot” in the title. Help can make a good action possible. It can’t make my refusal to participate count as having participated.
What We Lose When We Skip the Work
This gets uncomfortable because many of the things we do produce two results. There’s the visible output, and there’s what happens to us while we produce it.
A student writes an argument and discovers a contradiction in what she believes. A new manager prepares for a difficult conversation and realizes part of the problem is his own behavior. A parent listens long enough for a child to get past the easy answer. The task takes time, and something important develops inside that time.
We should be careful about removing the experience just because we can get a convincing version of the output without it. A well-written apology can help me find the words. If I’m unwilling to acknowledge what I did or change how I behave, the quality of the language won’t repair the relationship.
None of this makes difficulty sacred. Plenty of work is repetitive, exhausting, and badly designed, and removing it can give people room to think and attend to each other. At B:Side, we’re putting AI agents to work inside the organization I lead, so these questions follow me into staff meetings.
The responsibility is to understand what we’re removing. Does the tool help someone participate more fully, or does it make it easier to withdraw while a plausible result appears in their place? The answer depends on the activity, the person, and what the arrangement is meant to serve. Efficiency alone can’t settle it.
Someone Has to Make Room for the Beginner
For leaders, this gets especially concrete when we look at the people coming into our professions.
Earlier in my career, I advised young people to make themselves indispensable. Take on the routine work and become so useful that someone will invest in teaching you. (It worked for me, which is exactly the problem with the advice.) It reflected opportunities I’d received, but it gave too little credit to the people who made those opportunities possible. Someone noticed my effort. Someone let me attempt more than I knew how to do and corrected me when I needed it.
As routine assignments become easier to automate, we have to reconsider the bargain. A company can remove the work that once gave a junior employee a foothold and still expect experienced people to be available when it needs them. Across an entire profession, that expectation stops making sense. Where will the experience come from?
The book takes that obligation seriously. If human worth means anything, it should change the opportunities we make available, including time for instruction and room to make a correctable mistake. Telling a beginner that they matter while eliminating every route by which they can become capable is an incomplete answer.
There is a cost to teaching. There is also a cost to assuming someone else will always do it.
The Life the Tools Are Meant to Serve
These questions reach beyond work into friendship, family, art, and faith. My own faith shapes the book’s account of human life and its spiritual depth, and I also want readers to be able to examine the argument without agreeing with every conviction that brought me to it.
We can disagree about the deepest explanation of a person’s worth and still recognize the danger of making that worth depend on performance. Most of us already understand this when someone we love gets sick, grows old, or simply has a terrible day. We don’t need that person to demonstrate a competitive advantage before their life continues to matter.
What the Machine Cannot Do asks what it would mean to carry that understanding into the way we build institutions and use our tools. It’s a book about the life we’re trying to make possible as assistance becomes more powerful, and the obligations we still have to one another within it.
If you’ve been wrestling with these questions as a leader, a parent, a teacher, or someone trying to find your place in changing work, I hope you’ll buy a copy. Read it with one person or one responsibility in mind. The argument becomes useful when it reaches something you can actually do differently.
Buy What the Machine Cannot Do on Amazon
The machines will keep improving. We still have to decide what their improvement is for, and make room for the people whose lives are affected by our answer.