Trust and quality notes
- Last updated
- September 3, 2026
A job can end at five o'clock. Work does not. Children still need care, meals need cooking, relationships need attention, and people still seek challenge, dignity, and purpose. Paul Millerd argues that any serious account of life after automation must begin by expanding our definition of work beyond paid employment.[1]
His view is a useful correction to claims that AI will solve work altogether. Automation may change jobs, but it does not erase the activity through which people sustain a life and find meaning.
The source's central argument
Millerd says there is no true after automation unless society escapes what he calls a job-shaped reality distortion field.[1] Within that field, work is flattened into tasks attached to a job description and paycheck. People become numbers in discussions about layoffs, employment rates, weak links, and bundles of labor.
He contrasts that abstraction with his own days. He soothes a newborn early in the morning, takes a three-year-old to school, cooks, answers email, and finds time to write whether or not the writing earns money. Others may describe him as unemployed or say he is wasting his talents, yet he experiences a life full of work and fulfillment.[1]
This example grounds his challenge. Paid employment is one form of work, but care, household activity, and self-directed creation are also demanding and consequential. If automation reduces jobs, those forms do not disappear.
Millerd is skeptical of confident declarations that work will be solved within a few years. He notes that these claims often come from people intensely devoted to their own work, even when they regard their jobs as optional. They can change jobs or take time away while continuing to receive dignity, challenge, and purpose from lives organized around work.[1]
The confusion between jobs and work
Predictions about automation often begin by decomposing a job into tasks. That can help explain what software might perform. Millerd's objection is that the method can quietly shrink the subject. A human life is not a job description, and socially necessary activity is not limited to what appears in payroll data.
Care makes the gap impossible to ignore. A claim that nothing will be left to do sounds absurd beside the needs of a child. The work may be unpaid, but it requires patience, attention, and responsibility.
The distinction also changes how we interpret leisure. A free afternoon is not necessarily an empty space waiting to be filled by consumption. It may become time for care, making, friendship, learning, rest, or community. Millerd does not provide a policy program for this possibility. He argues that we lack language broad enough to imagine it.
Meaning does not arrive with a paycheck
Millerd identifies dignity, challenge, and purpose as things humans crave and can receive from work.[1] People in flexible positions may already understand this because their relationship to a job is optional while meaningful activity remains intense.
For someone whose job is not optional, talk about the end of work lands differently. Paid employment provides income and structure, and work outside the job continues after the shift ends. Announcing that work is solved can therefore ignore both economic dependence and unpaid responsibility.
This is not an argument that every job is fulfilling or that unpaid work is always freely chosen. The source does not romanticize all labor. Its point is that our categories are too narrow. A society planning for automation cannot treat a loss of employment as the removal of human purpose, nor can it assume purpose alone replaces income.
That final distinction is partly my interpretation. Millerd emphasizes the difference between jobs and work, but he does not develop an economic proposal in this short source. Readers should not infer a specific policy position that he has not stated here.
What is argument and what is interpretation
Millerd explicitly argues that modern discussion mistakenly equates work with jobs. He uses care, transportation, cooking, email, and writing to show that a full day can contain substantial work outside conventional employment. He also argues that people forecasting the end of work often remain deeply oriented around work themselves.[1]
My practical interpretation is that automation decisions should account for all the work surrounding a formal workflow. A tool that makes an office task faster may shift preparation, checking, explanation, or emotional burden onto someone else. If measurement notices only paid task time, that transferred work can remain invisible.
A second interpretation is that time saved should be discussed as a positive human resource. Teams often justify automation by counting output or cost. Millerd's perspective invites another question: what kind of life can the released time support? The answer may include better paid work, but it may also include caregiving, creative effort, or an unhurried afternoon.
These are applications of his expanded definition, not outcomes promised in the source. Automation can also increase demands or distribute free time unfairly. Millerd's essay gives us a lens for noticing those issues, not evidence that they will resolve themselves.
A broader audit of work
To use Millerd's idea, begin by listing everything required for a result, including activity outside the formal job. Who gathers information before the process begins? Who calms a worried customer, checks a confusing output, teaches a colleague, or handles the household work that makes paid concentration possible?
Then ask what automation removes, what it changes, and what it creates. A faster system may reduce one task while adding supervision. It may move effort from a specialist to a customer. It may save time at work but make a schedule less predictable. A broad account is more honest than a narrow task count.
Next, ask people what they want the saved time for. The question respects Millerd's insistence that a life contains many forms of work. It also prevents an organization from automatically filling every efficiency gain with more assigned activity.
Finally, protect language. Say job when you mean paid employment. Say task when you mean a discrete activity. Use work for the larger field only when you are prepared to include care, creation, and the maintenance of everyday life. Better terms do not solve material problems, but they make those problems harder to hide.
Life after automation is still full
Millerd's closing challenge is imaginative. Society finds it easier to picture catastrophe than a life with a free Tuesday afternoon.[1] That line of thought reveals how thoroughly the job has become the default container for purpose and time.
A future with fewer job tasks is not automatically a future without effort, need, or meaning. Humans will continue to care for one another and undertake difficult things. The open question is whether institutions will recognize that broader work and whether people gain real freedom to shape it.
The most useful response to automation is therefore not to declare work finished. It is to see more clearly how much work a human life already contains, which parts technology can lighten, and which parts deserve greater attention and respect.
If you want automation to return time to the life around the job, Agentic Workers can help you build a workflow around the work you actually want technology to carry.