Should the Government Have Protected the Luddites? And What Does the Answer Mean for AI?

The word Luddite has become an insult.

Call someone a Luddite today and you usually mean that they are afraid of technology, resistant to change, or unwilling to accept progress.

That is a convenient story.

It may also cause us to miss the most important question the Luddites left behind.

The Luddites were workers living through a technological transition that threatened their livelihoods, bargaining power, and economic value. They responded by attacking machinery.

That was not a viable way to stop industrialization, and destroying other people's property was not an acceptable economic policy.

But more than two centuries later, there is still a question buried underneath their resistance:

And that question may be much more important in the age of artificial intelligence than it was in 1811.

Should Their Jobs Have Been Protected?

At first glance, the answer might seem obvious.

No.

Government should not preserve every existing job forever simply because someone currently depends upon it.

If a machine can perform dangerous work more safely, we should probably use it.

If an excavator can dig in one afternoon what dozens of people once dug with shovels over several days, forcing society to keep using shovels merely to preserve employment would make little sense.

If automation allows food, clothing, transportation, medicine, or housing to be produced with less human effort, that productivity can benefit society enormously.

Industrialization ultimately did exactly that.

So perhaps government should not have protected the Luddites' particular jobs.

But that is not the end of the question.

Protect the Job—or Protect the Person?

A particular occupation may become obsolete.

A human being does not.

That distinction changes the discussion.

Suppose a worker has spent twenty years developing a skill that suddenly loses much of its economic value because of a new technology.

We could say: technology changes. Adapt.

There is some truth in that. People have always had to adapt to economic change.

But how much of that transition should fall entirely on the individual?

Should a fifty-year-old worker simply be expected to identify a new occupation, pay for the training, support a family while learning it, move if necessary, compete with younger entrants, and hope the new occupation itself does not soon disappear?

Modern societies already implicitly recognize that the burden cannot always fall entirely on the worker.

That is why unemployment insurance exists. It is why vocational retraining exists. It is why governments fund education and workforce-development programs.

Germany provides a useful example. Its Federal Employment Agency can support unemployed workers or people threatened with unemployment through vocational education and retraining. Under qualifying conditions, the agency can cover training costs and continue unemployment benefits while approved training is underway.

Germany is not necessarily prepared for widespread AI displacement. No country may be.

But the underlying principle is important:

A Layoff Is Not the Same Thing as a Labor-Market Transformation

There is another distinction we need to make.

A company laying off 500 workers is serious.

But it is not necessarily a systemic problem.

Perhaps the company lost a major customer. Perhaps management made poor decisions. Perhaps an industry is declining while another is expanding.

If other employers still desperately need the same skills, much of the problem is one of matching displaced workers with new opportunities.

The burden may reasonably fall significantly on the worker to search, relocate, learn something new, or accept a different position.

Now imagine something else.

Accounting firms need fewer junior accountants because AI performs much of their work.

Software companies need fewer junior programmers because AI writes increasing amounts of code.

Advertising firms need fewer copywriters and designers.

Law firms need fewer junior researchers.

Banks automate portions of analysis.

Insurers automate claims processing.

Customer-service departments become increasingly automated.

Administrative work contracts across thousands of companies simultaneously.

Eventually robotics begins reaching transportation, warehousing, manufacturing, construction, and service work.

That is not:

It is:

Those are fundamentally different problems.

An isolated layoff is primarily a labor-market event.

A widespread change in the demand for human labor is a systemic economic event.

And systemic problems cannot reasonably be assigned entirely to the individuals trapped inside them.

The Luddites Were Experiencing a Systemic Change

This is part of what makes their history relevant.

The Luddite disturbances began in Nottinghamshire in 1811 and spread into other textile regions. Britain's National Archives describes skilled textile workers opposing employers who used new machines to replace skilled labor and drive down wages. In different regions, the precise grievances varied, but unemployment, wage pressure, mechanization, and working conditions were all part of the conflict.

This was not merely one poorly managed textile company firing a handful of employees.

A technological change was altering the economics of an industry.

Machines affected who could perform work, how much output could be produced, what skills were valuable, how much bargaining power workers possessed, and what manufacturers needed from labor.

The Luddites understood something important even if their response ultimately failed:

That is much closer to the question AI presents today.

Technology Does Not Decide Who Pays

A machine does not decide what happens to the productivity gain it creates.

People and institutions do.

Suppose a new technology allows ten workers to produce what previously required twenty.

There are many possible outcomes.

Production might double. Prices might fall. Workers might earn higher wages. Workweeks might become shorter. Profits might increase. Demand might grow enough to preserve all twenty jobs.

Or the company might conclude that it now needs only ten workers.

Technology creates the possibility.

Economics, ownership, competition, bargaining power, and public policy help determine how the benefits and costs are distributed.

That is why technological progress is never purely technological.

It is also economic and political.

Follow the Interests

The different participants in a technological transition do not experience the transition in the same way.

For the manufacturer, new machinery may mean lower costs, greater output, and stronger competitiveness.

For the investor, it may mean higher returns.

For the consumer, it may mean a better or cheaper product.

For the worker, it may mean the destruction of a livelihood.

All of these people can be acting rationally.

They simply occupy different positions.

And this is where government becomes important.

A company can reasonably say: my responsibility is to remain competitive.

An investor can reasonably say: my responsibility is to seek a return.

A consumer can reasonably say: I want the best product at the lowest price.

A displaced worker can reasonably say: I need to support my family.

The problem is that there is no reason to assume that those individually rational positions will automatically produce a socially acceptable transition.

Government Does Not Have to Stop Progress to Govern Progress

This is where the debate often becomes too simplistic.

Either government stays out of the way and lets innovation proceed.

Or government interferes with technology and prevents progress.

Those are not the only choices.

Government can allow the machine.

And still govern the transition.

That requires more than political skill. It requires governing competence. Technological transitions like AI cut across economics, labor markets, education, science, ethics, taxation, social insurance, national competitiveness, and more. No elected official can master all of those fields, but government must have enough independent expertise—and enough intellectual humility—to understand when an industry, advocacy group, or ideological movement is presenting only part of the picture. Otherwise, the people with the strongest financial or political incentives can end up defining both the problem and the solution.

Government can provide education. Retraining. Income continuity. Mobility assistance. Apprenticeships. Temporary wage support. Healthcare continuity. Rules governing the speed or manner of deployment. Incentives to reduce working hours rather than immediately eliminate workers.

Or entirely new mechanisms we may eventually need if traditional employment itself becomes less central to income.

None of those ideas requires pretending obsolete jobs should remain forever.

They require recognizing that technological change can produce costs that individual workers did not create and cannot individually solve.

Retraining Works Only If There Is Somewhere to Retrain To

This is where AI creates a harder problem than many previous technological transitions.

Traditional retraining assumes something like this:

That can work extraordinarily well.

But what happens if AI is simultaneously entering A, B, C, D, and E?

What should the displaced copywriter retrain to become?

A programmer? AI is affecting programming.

An accountant? AI is affecting accounting.

A paralegal? AI is affecting legal research.

A graphic designer? AI is affecting design.

A financial analyst? AI is affecting analysis.

There will certainly be occupations that remain valuable.

There will almost certainly be new ones.

And many existing jobs may be augmented rather than eliminated.

But the old answer—just retrain people—becomes incomplete if the technological frontier is moving into many cognitive occupations at the same time.

Retraining works only if there is economically valuable work to retrain people for.

AI Did Not Sneak Up on Us

There is another reason government responsibility becomes harder to dismiss.

Artificial intelligence did not suddenly appear when ChatGPT became popular.

The roots stretch back generations.

In 1943, Warren McCulloch and Walter Pitts published an early mathematical model of neural networks that became foundational to later artificial-intelligence research.

In 1955, John McCarthy and other researchers proposed the Dartmouth Summer Research Project on Artificial Intelligence. The meeting took place in 1956 and helped establish AI as a formal field of study.

That was seventy years ago.

Governments cannot predict exactly what technologies will be capable of twenty years from now.

Neither can businesses. Neither can researchers.

But there is an enormous difference between failing to predict the exact future and failing to prepare for a clearly identifiable possibility.

We have known for generations that researchers were attempting to make machines perform increasingly sophisticated forms of human cognition.

Now the capability is visible.

Companies are openly discussing automation. Workers are using AI every day. Researchers measure which occupations are exposed. Investors are placing enormous bets on AI replacing or augmenting human work.

If widespread labor disruption eventually arrives, it would be difficult to characterize it as an unforeseeable event.

That Changes the Government's Responsibility

A sudden factory fire is unpredictable.

A technological transformation unfolding over generations is not the same thing.

Government does not need to know exactly what will happen.

It needs to ask:

What happens if displacement is modest?

What happens if it is severe?

Which occupations are most vulnerable?

How quickly can workers realistically move?

What happens to healthcare when employment falls?

What happens to retirement contributions?

What happens to tax revenue?

What happens to housing?

What happens to aggregate consumer demand?

What happens if retraining programs produce workers for occupations AI is simultaneously reducing?

The People Who Benefit Often Have More Power to Define Progress

There is another uncomfortable lesson from the Luddite era.

The people gaining from technological change often possess more power than the individuals absorbing its immediate costs.

That does not require villains.

The factory owner had capital. The worker had labor. The government controlled law and force.

Those positions were not equivalent.

Today the distribution of power is different, but the asymmetry remains.

Technology companies can spend billions developing AI.

Large corporations decide how it will be deployed.

Investors influence capital allocation.

Governments determine policy.

Industry groups lobby.

Researchers influence the direction of technology.

The individual worker whose occupation begins disappearing has comparatively little influence over any of those decisions.

The issue is not that everyone with power is bad.

Many are not.

It is that the people deciding how quickly a transformation proceeds may not be the same people who bear its greatest downside.

That matters.

The Luddites Were Ultimately Defeated

Industrialization continued.

And it ultimately produced extraordinary benefits.

Goods became more abundant. Productivity increased. New industries appeared. Living standards ultimately rose enormously.

That history matters.

It means that simply stopping technological progress would likely have imposed huge costs of its own.

But there is a logical mistake hidden inside the happy ending.

Those are separate questions.

A technological transformation can be good for civilization over fifty years while being catastrophic for particular people over five.

We tend to remember the eventual prosperity.

The people who absorbed the transition cost often disappear into history.

So Should the Government Have Protected the Luddites?

If by protected we mean: should Britain have permanently forbidden labor-saving machinery so textile jobs could never change?

Probably not.

But if we ask:

The answer is far less comfortable.

Perhaps the failure was not allowing machines to exist.

Perhaps the failure was allowing the question to become: machines or workers.

A more competent society might have asked:

That is the question the Luddites still leave us.

And Now We Have AI

Artificial intelligence may produce astonishing prosperity.

It may complement human labor more than it replaces it.

It may create jobs we cannot yet imagine.

It may allow shorter workweeks and dramatically higher living standards.

That future is possible.

But another future is possible too.

AI could reduce the amount of human labor required across many industries at roughly the same time.

If that happens, telling each displaced worker individually to adapt will not constitute an economic policy.

It will be society refusing to acknowledge a systemic change.

The question therefore is not simply:

The better question is:

We did not answer that question particularly well for the Luddites.

We have had more than two centuries to think about it since.

And we have known for generations that machines were becoming capable of more than physical work.

This time, if we wait until the disruption arrives before deciding what comes next, we will not be able to say we had no warning.