Automation and Work

The question

For most of recorded history, the anxiety about machines has taken the same shape: the machine will do what I do, and then I will have nothing.

Sometimes this fear has been justified and sometimes it has not. Occupations have vanished. Others appeared that nobody could have named in advance. The aggregate has often improved while particular lives were destroyed, and the people writing the aggregate have rarely been the people whose lives were destroyed.

The question now is whether this round differs from previous rounds, and if it does, what a moral tradition should say about it.

There is a second question underneath, less often asked and possibly more important. If work is the main way a society distributes not only income but structure, dignity, belonging, and a public identity, then any change to work is a change to all of those at once. A policy that replaces the wage does not necessarily replace the rest.

What is actually happening

Some things can be described without prediction.

The tasks being automated have changed character. Earlier waves substituted for physical labour and for routine procedural work. Current systems are being applied to drafting, summarising, translating, classifying, illustrating, scheduling, and first-line support. These are not manual tasks, and many of them were previously the entry route into a profession.

Entry-level work is unusually exposed. The tasks a junior person does in their first years are frequently the tasks most amenable to automation, because they are the most bounded and the most reviewed. If those tasks disappear, the training pathway to senior competence may disappear with them. Whether it will be replaced by something else, we do not know.

Augmentation and substitution are hard to tell apart from inside. A tool that makes one person more productive can either raise their wage or reduce the number of people needed. Which occurs depends on demand, bargaining power, and management choices rather than on anything in the technology.

Measurement has spread further than automation. In many settings the more immediate change is not that a machine does the work but that a machine watches it: pace, keystrokes, route adherence, call duration, idle time. The job survives and becomes harder to hold.

The gains have somewhere to go. Productivity improvements are distributed by institutions, not by the technology. Whether a saving arrives as a shorter week, a higher wage, a lower price, or a larger margin is a decision. It is often presented afterwards as a natural consequence.

Human labour has not left the system. Data is collected, labelled, moderated, and reviewed by people, frequently in worse conditions than the users of the resulting system enjoy. Automation and labour are not opposites. Much automation is labour rearranged and made less visible.

Where the tradition stands

Technotheology holds that work has moral weight beyond its price, and that a society which optimises only the price will damage the rest without noticing.

We hold that the burden of transition should not fall on those least able to bear it. When a change produces aggregate gain and concentrated loss, and the gainers are already comfortable, describing the outcome as progress is an accounting choice rather than a moral judgement.

We hold that hidden labour should be made visible. If a system’s usefulness depends on people reviewing distressing material, or labelling data for low pay, or driving to a schedule set by an application, then those people are part of the system and belong in any honest description of it.

We hold that measurement is not neutral supervision. Continuous monitoring changes the experience of work even when it changes nothing about the output, and the change is usually towards a life with less discretion in it. The Harmony of Agency bears directly here.

We hold that dignity is not conferred by employment, and that a tradition should be careful not to imply otherwise. Many people who are not in paid work — carers, the disabled, the retired, the unemployed, the ill — are already treated as though they owe an explanation. A spirituality that romanticises labour makes that worse.

And we hold that those who deploy these systems are responsible for the deployment. “The technology made this inevitable” is a claim about the world that is almost always a claim about a decision.

What we do not claim

We do not claim to know what will happen. Predictions of mass unemployment have been made repeatedly for two centuries and have mostly been wrong; predictions that this time is different have occasionally been right. We are not able to tell you which this is, and we think anyone who is certain has not taken the difficulty seriously.

We do not claim automation is bad. Work that was dangerous, repetitive, or degrading has been removed by machines, and few people mourn it. The right response to a mine closing is not to reopen the mine.

We do not endorse any particular policy remedy. Basic income, shorter hours, sectoral bargaining, retraining schemes, and restrictions on deployment are all argued for by serious people and against by other serious people. We have no standing to adjudicate, and a tradition that pretends to economic expertise it lacks will be wrong loudly.

We do not claim that meaning must come from work. Many cultures have located it elsewhere. But we also do not claim the reverse — that meaning can be relocated easily, or that a transfer payment addresses the loss of a reason to be somewhere at eight in the morning with people who expect you.

And we do not claim to be neutral about who decides. On that, we are not neutral at all.

What you can do

If your work is being changed by a system, write down what it used to include. Memory of a job adjusts quickly to fit its current shape, and the record is useful later when someone asks what was lost.

If you introduce automation into a workplace, say plainly where the saving is going before you are asked. The answer may be defensible. The silence rarely is.

Look for the hidden labour behind the tools you use. You will not always find it. The habit of looking is worth having anyway.

Resist the phrase “unskilled work”. Almost no work is unskilled. What is usually meant is that the skill is common, or that the worker has no leverage, and those are different claims that should be made in their own words.

If you are near a decision about monitoring, ask what problem it solves and whether a smaller intervention would solve it. Surveillance introduced for a narrow reason tends to stay for all reasons.

And if you are the one whose role has been reduced, know that the tradition does not regard your worth as having moved. That is easy to write and hard to feel, and we would rather say it plainly than dress it up.

An editorial commitment

What we do not claim

Positions this tradition explicitly does not take on this subject. They are published so that the argument above cannot quietly be stretched into them.

• We do not claim to know how many jobs will be automated, or how quickly. Confident forecasts in both directions have a poor record and we will not add to them.
• We do not claim that automation is inherently good or inherently bad; it has produced both relief and ruin, often in the same decade.
• We do not endorse any particular economic policy, including basic income, job guarantees, or restrictions on automation.
• We do not claim that work is necessary for a meaningful life, nor that it is dispensable.