Tasks, Not Jobs

Technology automates tasks, not whole jobs — so work doesn't vanish, it reshapes, and what's scarce and valuable keeps shifting.

The Future of Technology · Lesson 16 · 8 min read

Every wave of technology brings the same headline — “the machines are coming for the jobs” — and yet, century after century, there is still plenty of work, just different work. But this time skeptics say it is different, because AI does thinking. Who is right — and what is the actually-useful way to think about technology and jobs? Hold that question; the answer is sharper than either panic or denial.

Technology automates tasks, not whole jobs

Almost every job is a bundle of many different tasks. A new tool usually automates some of those tasks, which changes the job — and often makes the worker more productive on the rest — rather than deleting the whole role.

So “will AI take this job?” is usually the wrong question. The sharper one is “which tasks does it change, and what is left for the human?” A tool that automates three tasks out of ten reshapes a job; it does not erase it.

What is scarce shifts — that is the real story

When a task becomes automated and cheap, value moves to whatever is still scarce and now complements it (the scarce-complement idea from Lesson 13). Historically that has marched from muscle, to operating machines, to judgment, creativity, relationships, and care.

Each wave did not end work — it moved what work is worth most, and it created entirely new categories no one could name in advance (Lesson 14).

ATMs and bank tellers: the obvious first-order guess was “machines dispense cash, so tellers vanish.” What actually tended to happen over decades — automating cash handling made a branch cheaper to run, so banks opened more branches, and tellers shifted toward the tasks machines could not do: helping customers, selling services, solving problems. The job changed; it did not simply disappear.

Honest about the transition

The grounded view avoids both fairy tale and doom. There is no law guaranteeing that every displaced person smoothly finds equal new work — transitions are uneven and can be genuinely painful for specific people, skills, and places (a sharp second-order effect, Lesson 15). And “this time AI does thinking” is a real difference worth taking seriously, not waving away.

But the pattern is strong: automation reshapes the mix of work and what we value rather than abolishing work. So the real question is not “will there be work?” but “how do we manage the adjustment well and help people move to where the new value is?”

Apply tasks-not-jobs to a worker today whose role includes ten tasks, three of which a new AI tool can do:
1. The three automatable tasks get faster and cheaper — the tool handles them.
2. The worker now spends more time on the other seven, and can handle more work overall — more productive, not unemployed.
3. Value shifts toward the seven tasks the tool cannot do — judgment, dealing with people, deciding what matters — which become the scarce, well-paid part of the role.
4. New tasks appear (checking and directing the tool) that did not exist before.
The first-order guess (“the job is gone”) misses that a bundle minus three tasks is a reshaped job, not no job — though if the worker’s value was mostly in those three tasks, the transition for them is real and hard, which is the honest part.