Full article · 7 min read

AI and Jobs: Will Automation Take Work Away or Create Something New?

Artificial intelligence has become one of the biggest flashpoints in the long-running debate about technology and work. New tools promise greater productivity, faster decision-making, and the ability to automate tasks that once required human labor. But they also raise a deeply personal question: if machines can do more, what happens to people?

That question is not new. Across history, technology has repeatedly changed how humans work. Since the invention of the wheel, technologies have helped increase economic output. Machines and systems have often replaced people in some lower-paying jobs, while also helping create new kinds of work. What makes AI feel different is the scale of its ambition. Unlike older tools that mainly assisted with physical labor or narrow calculations, AI is often discussed as a system that could perform a much wider range of tasks.

A Forecast That Fuels the Debate

One widely cited prediction comes from the World Economic Forum’s "The Future of Jobs Report 2020." It says AI is expected to replace 85 million jobs worldwide and create 97 million new jobs by 2025.

That forecast captures why the AI-and-jobs debate is so difficult. It does not describe a simple story of collapse or triumph. Instead, it points to disruption: jobs disappearing in some places, jobs emerging in others, and a workforce forced to adapt in between.

In plain terms, replacing 85 million jobs means certain roles may no longer need as many human workers because software, algorithms, or machines can handle parts of the work. Creating 97 million jobs means new roles, industries, and support systems may arise around these same technologies. The headline numbers sound dramatic because they are dramatic, but they also suggest transformation rather than a one-way march toward joblessness.

What History Says About Technology and Employment

Looking backward helps explain why many economists resist simple predictions. Past automation has both substituted for labor and complemented it.

To substitute for labor means a machine takes over a task a person used to do. A classic pattern is technology replacing humans in some lower-paying jobs, including in agriculture. To complement labor means technology makes people more productive, often changing their jobs instead of eliminating them outright.

This mixed pattern matters because it shows that technological change has rarely been all good or all bad for workers. Technology is described as the largest cause of long-term economic growth, and it has contributed to increased prosperity, improved comfort and quality of life, and medical progress. At the same time, it can disrupt existing social hierarchies and harm individuals or groups.

That tension sits at the center of the current AI discussion. Supporters see another wave of productivity and economic growth. Critics see a risk that the gains may not be shared evenly, especially if some workers lose jobs or face lower wages while others benefit.

Computers Didn’t Cause Significant Net Unemployment

One reason some analysts are cautious about AI doom scenarios is that earlier digital technologies did not produce a clear long-term collapse in employment. Studies have found that computers did not create significant net technological unemployment.

Technological unemployment is the idea that workers can lose jobs because machines or automated systems do the work instead. The concern has been around for a very long time. It is not unique to AI, or even to computers.

This historical point is important because it tempers extreme claims. If past computing revolutions did not result in significant net unemployment, then it is possible that economies can adapt to major technological shifts. New jobs can emerge, industries can reorganize, and workers can move into different kinds of roles.

But that is only part of the story.

Why AI May Be Different

The biggest reason there is still no clear answer is simple: AI may not behave like previous waves of technology.

The article notes that artificial intelligence is far more capable than computers, and still in its infancy. “In its infancy” means it is still in an early stage of development, not yet fully mature or fully understood in terms of long-term effects. Because of that, it is not known whether AI will follow the same trend as earlier computerization.

That uncertainty is exactly why economists and policymakers continue to debate the issue at length. A 2017 survey found no clear consensus among economists on whether AI would increase long-term unemployment.

A lack of consensus means experts do not agree on the outcome. Some may expect adaptation and job creation. Others may expect deeper labor market disruption. And many likely believe the result depends on how AI is developed, adopted, regulated, and distributed across industries.

What Robots Have Already Done to Labor Markets

While the future remains unclear, there is already evidence that automation can hurt some workers in the present.

From 1990 to 2007, a U.S. study by MIT economist Daron Acemoglu found that adding one robot for every 1,000 workers decreased the employment-to-population ratio by 0.2%, or about 3.3 workers, and lowered wages by 0.42%.

The employment-to-population ratio is the share of people of working age who have jobs. So when that ratio falls, it means a smaller share of the population is employed.

This finding matters because it gives the debate real-world grounding. It shows that automation does not merely reshuffle abstract economic statistics. It can affect whether people work at all, and how much they earn.

The wage drop may sound small, but across large populations it can be significant. And the robot effect also highlights a recurring pattern in technology: benefits and burdens are often unevenly distributed. A company may become more productive, consumers may enjoy cheaper goods or faster services, and the economy may grow overall, while certain workers or communities still lose out.

Productivity, Progress, and the Human Cost

Technology is often defended because it increases productivity. Productivity means producing more output with the same or fewer inputs, such as labor, time, or energy. Higher productivity is closely tied to economic development, and throughout history new technologies have expanded what societies could do.

But productivity gains do not automatically guarantee fairness.

That is why concern about machines replacing workers has endured for generations. When President Lyndon Johnson signed the National Commission on Technology, Automation, and Economic Progress bill in 1964, he said: “Technology is creating both new opportunities and new obligations for us, opportunity for greater productivity and progress; obligation to be sure that no workingman, no family must pay an unjust price for progress.”

That sentence still feels strikingly modern. It recognizes both sides of the technological bargain: more output and more possibility, but also a responsibility to deal with the social costs.

Why the AI Debate Is Still Open

The most honest answer to “Will AI erase jobs?” is that nobody knows for sure.

There are good reasons for optimism. Technology has repeatedly helped raise economic output. Past computers did not lead to significant net technological unemployment. Forecasts suggest AI could create even more jobs than it replaces.

There are also good reasons for concern. AI is more capable than previous computer systems, and its development is still at an early stage. Evidence from robotics shows measurable negative effects on employment and wages in some settings. Economists themselves do not agree on the long-term outcome.

This makes AI less like a settled story and more like an unfolding experiment. The result may depend on where automation is introduced, which tasks are targeted, how quickly workers can transition, and whether the gains from AI are broadly shared.

Technology Has Always Changed Work

Seen in the widest historical frame, AI is part of a much older pattern. Technology has transformed human life from stone tools and fire to the wheel, printing press, telephone, and internet. It has lowered barriers to communication, improved prosperity, and reshaped the structure of society. But it has also produced pollution, social disruption, and fears about labor displacement.

AI fits squarely into that tradition. It is another powerful example of conceptual knowledge being applied to practical goals in a reproducible way. The question is not only whether it can do more work, but what kind of society that new capacity will create.

The future of jobs may not be decided by technology alone. Modern thinking increasingly looks at sociotechnical systems: combinations of things, people, practices, and meanings. In other words, tools matter, but so do institutions, policies, values, and choices.

That may be the clearest lesson of all. AI is changing the world of work, but the final outcome is not written yet.