Will AI trigger mass unemployment quickly? Will it overturn entire industries? And how should ordinary people respond?

As anxiety over the impact of artificial intelligence spreads across the world, National Business Daily (NBD) recently interviewed Peter Howitt, winner of the 2025 Nobel Prize in Economic Sciences, on how he views the AI wave through the lens of technological change.

As an economist known for his work on innovation and growth, Howitt — together with Philippe Aghion — helped build the mathematical framework of creative destruction, the idea that new innovation replaces old technologies and in doing so drives economic development forward.

From that perspective, Howitt argues that it is still too early to say AI will broadly replace human beings. But he also says people should be ready for multiple job changes over the course of their lives. In his view, the deeper long-term risk is not simply job replacement, but the possibility that wealth and power become concentrated in the hands of a small number of companies. To address that, he says, a wealth tax would make more sense than a robot tax.

Below are edited excerpts from the conversation.

Peter Howitt (Left) Photo/VCG

The computer revolution empowered top talent and AI revolution may help ordinary workers more

NBD: How would you define the current AI wave? Do you see it as a new round of creative destruction?

Peter Howitt: First, you have to recognize that generative AI is very new. ChatGPT was the first generative chatbot, and it is only a little over three years old.

We still do not know whether large language models will ultimately be the future of artificial intelligence. It may instead be something more like the “world model” approach that Yann LeCun has been developing. It may also turn out to be a more modular approach, such as the one we have seen in China, where DeepSeek has been very successful.

So there is still a great deal of uncertainty. Anything I say can only be suggestive. Nobody has a crystal ball.

That said, AI does share many characteristics with what we would call a general-purpose technology. It is obviously going to have a revolutionary effect on the way things are done throughout the economy. In that respect, it is not very different from the automobile.

Before automobiles became widespread, most people born in North America never ventured more than a few miles from where they were born. Now, with the automobile, jet aircraft, and so on, even people of average income are able to travel around the world. That has had an extraordinary effect on people’s lives.

So yes, AI is another such technology. Whether it will prove more or less revolutionary than earlier ones, I do not think we know yet. It obviously has tremendous potential.

But I do think it differs from the computer revolution in one important way. The computer revolution was very much a skill-biased technological change: the more skilled you were, the more it raised your productivity.

By contrast, AI appears, in general, to have a negative skill bias. It seems mainly useful for automating routine cognitive tasks, which can be very helpful for people who are skilled, but not necessarily at the very top of the skill ladder.

The further up the ladder you go — say, to the CEO of a major corporation — the more you already have other people taking care of many of your routine cognitive tasks. But think of someone like a nurse. A nurse may be able to make tremendous use of artificial intelligence in providing services, by drawing on all sorts of information available online that is currently time-consuming to access. And if AI frees up her time from routine data entry, for example, she can spend more of it dealing with patients.

So in that sense, AI may be somewhat different from the computer revolution.

AI is already taking some jobs — but it will also create new ones, many of them in services

Peter Howitt (Right) Photo/Nobel Prize website

NBD: AI has already become a reason cited by some tech companies for layoffs. How do you see the risk of job displacement?

Peter Howitt: Every major revolutionary technology makes some human skills obsolete. When the automobile arrived, it greatly reduced the demand for blacksmiths and people who trained horses, because horses were no longer needed as much for transportation.

Something similar is going to happen here, and we can already see it beginning to happen in software coding. There is clearly a decline in demand for software coders. How much of that is due to AI and how much is due to past overinvestment in coding, I do not know. But tools like Claude Code have clearly changed the market for coders because they can do so much. All you have to do is give them instructions.

That said, the people operating systems like Claude Code still need to know a fair bit about coding in order to give good instructions. So I think we are still going to have to teach people how to code.

The job market is clearly going to change. Companies will be generating many more lines of code than in the past, but those lines of code will still need to be checked and verified. We will need supervision, and that supervision will require people who are expert in coding. So some jobs will remain.

At the same time, any general-purpose technology also opens up new occupations and new skills — typically occupations and skills that nobody has even heard of before, because the new technology opens up new possibilities.

A revolutionary technology like this does not simply replace humans with machines; it also reorganizes the way everything is done. And that reorganization opens up new skills and occupations.

I often tell people to imagine themselves back in North America in 1875, when it took 50 percent of the population working on farms, doing very difficult labor, just to feed everybody. Now imagine telling those people that within 150 years it would take only a little more than 1 percent of the population to do that work, and that the other 49 percent would no longer be needed on the farm.

They might ask: what are all these people going to do? And the real answer would be: some of them will be bloggers, some of them will fly jet planes, and so on. But none of that would have made any sense to someone in 1875, because those occupations did not yet exist. They depended on technologies that would have seemed like science fiction at the time.

That, too, is likely to be true with AI.

NBD: Will white-collar and middle-class workers face a bigger shock?

Peter Howitt: It is still too early to say anything with confidence. But in general, as I said, I do think there is a negative skill bias to this technology.

And I think the workers whose productivity will be enhanced the most are people who are part of the middle class. That is not to say there will be increased demand for all of their services. What I can imagine happening in many professions is that AI will automate many routine tasks and make workers much more productive.

In particular, it may make the most productive workers in a given field so productive that they are able to command much more of the market, leaving room for fewer people overall. As a result, many people may lose their jobs. Demand for their services may decline because demand for the very most productive people in the field has risen so much.

Where those displaced workers will go is not yet clear. If the economy were completely self-adjusting — if Say’s Law were fully at work — there would always be employment for them. But exactly where that employment will come from will depend on downstream innovation.

I also think it is easy to exaggerate how quickly all of this will happen. The technology is evolving very rapidly, but the ability to replace large numbers of workers with bots of various kinds is still fairly limited.

There are not many things you could completely entrust to AI right now. It makes mistakes. It hallucinates. People thought they might replace legal workers with bots, but these bots sometimes generate beautiful-looking briefs that could be presented to a judge, except that they refer to precedents they invented and that do not actually exist. If you do that in front of a judge, you are in serious trouble.

Most people’s jobs still involve a great deal of human judgment, imagination, creativity, person-to-person contact, empathy and persuasion — qualities that remain uniquely human.

That is where tacit knowledge matters. Tacit knowledge is the kind of knowledge you acquire through experience, intuition and observation, often in ways that you cannot fully write down. A great deal of what people do, even in highly cognitive jobs, involves tacit knowledge. Most jobs still require things AI is not very good at. So it is going to take a long time before many of these jobs are replaced.

And in the meantime, new jobs and new occupations are going to emerge. Historically, that has been the pattern. What is most obvious at first is the set of jobs that are clearly at risk. What you cannot see yet are the jobs of the future, because those depend on future innovations.

So yes, I am sure it will be a bumpy ride for a lot of people. But I do not think it is going to happen as fast as many people fear.

Many of the new jobs AI opens up will likely be in the service sector. I do not think there is any stopping the continued decline in labor demand in manufacturing.

Photo/ Brown University website

AI could concentrate wealth further and a wealth tax makes more sense than a robot tax

NBD: Some people worry that AI will concentrate more wealth and power in the hands of a small number of companies. How do you see that risk?

Peter Howitt: Yes, I think that is definitely a risk. But it is a controllable risk.

Governments can exert a great deal of leverage over companies. They can regulate how companies operate, and they can use tax systems to limit the extent to which financial power becomes concentrated among AI developers.

If governments simply take a hands-off approach and say, “Whoever develops the best system can run with it,” without becoming directly involved — which, to a large extent, is what is happening in the United States — then yes, there is a big risk.

NBD: OpenAI recently suggested that governments could set up a public fund or impose a robot tax so that ordinary people can share in the gains created by AI. Do you think that idea would work?

Peter Howitt: I am not sure a robot tax is the best way to go. I think you need a broader-based tax, rather than one focused narrowly on robots. Companies will always find ways around a narrow tax, and that is not necessarily efficient for society; it is only efficient for the companies trying to avoid it.

The broader the tax base, the smaller the distortions in incentives.

What I think you could do is introduce a wealth tax, which in my view is long overdue in most countries anyway.

In most countries, the main source of government revenue is the income tax. But we define income in a way that excludes unrealized capital gains. That is a recipe for wealth concentration, because people continue to build wealth and the growth of that wealth itself is not taxed.

And when wealthy people want to draw on that wealth to finance consumption, they often do not sell the assets and realize capital gains. Instead, they borrow against their wealth. In the United States, if people become wealthy enough, they can just keep doing this indefinitely. Then, when they die, they pass their wealth on to their children. And in practice, that wealth largely escapes income taxation altogether because there is effectively very little inheritance tax.

That is one reason such enormous fortunes can accumulate. And I think a wealth tax is in order to redistribute some of those very large gains more broadly across the population.

NBD: Beyond that, what about universal basic income, the UBI?

Peter Howitt: I do think that at some stage, in advanced countries like the United States, a UBI could make sense. But I do not think we are there yet.

Before that, I think we need a wealth tax. Because if a UBI is financed only through income taxes, there simply is not enough revenue to make it substantial for everyone. It would have to be a very small payment, unless income tax rates were raised to implausibly high levels.

But with even a 2 or 3 percent wealth tax, you could generate much more revenue, and that might make a meaningful UBI possible.

Ordinary people should prepare for multiple career shifts over a lifetime

NBD: What would you say to ordinary people, especially workers who feel anxious or under pressure because of AI?

Peter Howitt: First of all, artificial intelligence is clearly going to become something everyone will need to use in order to obtain gainful employment in the future. So people really do have to embrace it and learn how to use it as effectively as possible.

Second, if history is any guide, there are likely to be even better opportunities in the future than there have been in the past. That does not mean the transition will be smooth. It will involve a great deal of job change. As technology evolves rapidly, people will increasingly need to be prepared for multiple career changes over the course of their lives.

Life keeps getting better. We take for granted things like automobiles, televisions and mobile phones, all of which have greatly improved our lives. But every one of those technologies also created tremendous uncertainty at the time, because they automated some jobs and disrupted others. That is happening again.

People got through it in the past, and they will get through it in the future.

There is no economic development — for an entire country, let alone for the world as a whole — that takes place without disruption, without job loss and job creation, without uncertainty and anxiety.

So you need to be optimistic. And you have good reason to be optimistic. But you should not expect a smooth ride.

If I were young, I would be looking forward to a world in which many of my efforts might end in failure, but if I persisted, I would be better off in the future than I am today.

The jobs that will be created are almost certainly going to be more interesting, more challenging and more rewarding than the jobs that are being destroyed.

So I would simply say: be hopeful, and be optimistic.

Editor’s note: This interview has been edited for clarity and concision. Portions of the original wording have been revised for editorial purposes without changing the intended meaning.