Economists are changing their minds about Artificial Intelligence and jobs
Artificial Intelligence has always been one of those topics that split opinions in the world of economics. On one side, Silicon Valley enthusiasts swearing everything would change overnight. On the other, serious economists pouring cold water on the hype and saying history had already played this movie before — and that employment always survived.
But something has changed.
It didn’t happen all at once, but it was noticeable enough to make headlines in the New York Times: economists are starting to take the impact of AI on the labor market seriously. And look, it’s not because they became die-hard fans of the technology — it’s because the signals are getting too hard to ignore. 👀
Daniel Rock, an economist at the University of Pennsylvania who studies the economic impact of artificial intelligence, summed up the current moment well: he believes AI hasn’t hit the labor market full force yet and hasn’t radically changed business productivity either, but that it’s coming.
The big concern right now isn’t just about unemployment itself, but about something far more pressing:
- Public policies aren’t ready for what may be coming
- The economy hasn’t shown the numbers yet, but early signals are already showing up around the edges
- And the window to prepare exists — but it won’t stay open forever
In this article, we dive into what the experts are saying now, what changed their thinking, and what it all means in practice for anyone who works, studies, or simply wants to understand what’s ahead. 🚀
What changed in the minds of economists?
For years, the consensus among economists was almost unanimous: technology creates more jobs than it destroys. The logic was simple and backed by history. The Industrial Revolution scared everyone in the 19th century, but the world of work reinvented itself. The rise of computers in the 80s and 90s generated massive fear about the end of entire professions — and the opposite happened: roles nobody had imagined before emerged. So why would this time be any different? That was the question that kept the optimism alive for a long time.
Even rising unemployment among recent graduates was frequently attributed to high interest rates and macroeconomic uncertainty. Predictions of massive job losses were dismissed as failures to learn from history’s lessons. And when companies laid off workers and blamed AI, many economists labeled it a form of AI washing — executives using the technology as a scapegoat to cover up poor management.
The problem is that today’s Artificial Intelligence doesn’t work like previous technologies. While the machines of the past replaced physical, repetitive tasks — freeing humans for cognitive and creative functions — current AI is moving in exactly the opposite direction. It’s advancing first into cognitive tasks, the ones that require reasoning, language, analysis, and content creation. This represents a fundamental shift that traditional economic models simply weren’t built to process, and the most honest researchers are admitting that openly now.
A recently published study by a team of researchers — including Ezra Karger, an economist at the Federal Reserve Bank of Chicago — surveyed economists on their outlook for the next five and 25 years. Most expect the economy to grow a bit faster as AI advances, but without straying too far from historical patterns. However, if the technology advances at an accelerated pace — something they consider unlikely but plausible — the picture changes dramatically: faster growth, but also greater inequality and the disappearance of millions of jobs.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
The most interesting part is that the economists’ expectations for the future turned out to be relatively similar to those of professionals from the AI industry itself, who were also surveyed for the study. Both groups agree that the future is uncertain: AI could either eliminate entire categories of work or cause very few losses. The effects could concentrate on entry-level office workers or spread to more experienced professionals and manual labor sectors. And the changes could turn the economy upside down in a few years or take decades to fully materialize.
The paradigm shift that convinced the skeptics
Alex Imas, an economist at the University of Chicago, is a good example of this transformation in perspective. When OpenAI released ChatGPT to the public in November 2022, he didn’t necessarily see it as an economic game-changer. The technology was powerful but limited, prone to errors, and incapable of producing work with the quality and consistency needed for most professional applications.
For Imas, the real turning point came at the end of 2024, when OpenAI released a model capable of reasoning — meaning it could work through a problem step by step before producing an answer. This capability enormously expanded the types of problems the model could solve and made its responses far more reliable. In his words, it was a paradigm shift, and he began thinking it was potentially an event on the scale of the Industrial Revolution, if not bigger.
For other economists, the penny dropped more recently, with the launch of Claude Code — a tool from AI company Anthropic that writes computer code from user prompts — and the spread of so-called AI agents, autonomous systems capable of executing tasks directly, without constant human supervision.
Molly Kinder, a senior fellow at the Brookings Institution who studies AI, shared that when she tried the new tools, she had a revelation: she simply no longer needed anyone to do the kind of basic research she normally hired college students and recent graduates to perform — and that she herself had done early in her career. According to her, if someone can do their job locked in a closet with a computer, that person is eventually going to have problems. More senior positions that require client and investor interaction or strategic decision-making may be safe for now. But the message for entry-level roles is clear.
The first signs in the labor market
The economy hasn’t yet produced catastrophic numbers about unemployment directly caused by Artificial Intelligence. But the first signs are appearing in specific sectors, and they’re clear enough to raise concern among those studying the issue closely.
Martha Gimbel, executive director of the Budget Lab at Yale University, makes an important counterpoint: technological advancement alone doesn’t reshape the economy. For that to happen, companies need to adopt the tools and figure out how to use them productively. And history shows that process almost always takes longer than the inventors expect. Legal and regulatory barriers slow things down. Companies need to retrain employees or hire new ones. Corporate leaders need to develop new processes and overcome resistance from reluctant managers and cautious IT departments.
Gimbel pointed out that many hospitals kept patient records on paper for decades after the technology to digitize them already existed. Video conferencing tools had been around for years, but it took a pandemic to force companies to actually adopt them.
Even so, there are signs that AI could spread through the economy faster than previous innovations. Data from the U.S. Census Bureau shows that nearly one in five businesses reports having used AI in the last two weeks, and in some sectors the rate is double that. Individual workers report using AI at even higher rates, suggesting many are experimenting with the tools on their own.
And although AI hasn’t yet generated a massive impact on aggregate statistics, some economists argue that its effects are already visible beneath the surface. Researchers at Stanford University published a study showing that employment was declining for entry-level workers in roles highly exposed to AI. Erik Brynjolfsson, one of the study’s authors, acknowledges that technological advances sometimes take decades to show up in the economy in the form of productivity gains — but believes this time it won’t take decades.
Speed and scope make all the difference
What makes this moment different from other technological transitions is the speed. When the internet transformed the labor market in the 90s and 2000s, the process took decades to reach global scale. The adoption curve for generative AI has been far steeper: in less than two years after the public launch of tools like ChatGPT, companies worldwide were already integrating these solutions into their daily workflows.
Predictions coming from Silicon Valley are far more intense than those from economists. Dario Amodei, CEO of Anthropic, warned that AI could eliminate 50% of entry-level office jobs within a few years. Tech investor Vinod Khosla predicted that AI would replace 80% of jobs by 2030. And Elon Musk went so far as to say the technology would make work optional.
Many economists dismiss such extreme predictions, arguing that the AI debate should focus less on where the economy will end up and more on the potentially painful transition period. Martha Gimbel from Yale’s Budget Lab framed the central question directly: the urgent question is about how painful the technology shock is going to be.
According to some estimates, up to 70% of jobs are in some way exposed to AI. But that doesn’t mean all those workers are about to be laid off. A Boston Consulting Group report published recently estimated that more than half of U.S. jobs will be reshaped by artificial intelligence over the next two to three years, but far fewer will be fully replaced. Most workers perform a range of tasks, and only some of them can be reliably handled by AI. On top of that, even where it would be possible to replace a worker, companies are moving cautiously because the risks increase when humans are no longer overseeing the machine’s work.
Greg Emerson, the report’s lead author, summed it up: full job replacement is happening much, much more slowly because implementation is harder, while the augmentation and reshaping of roles is happening much, much faster.
The combination of speed and scope is what will determine how smooth or traumatic this transition turns out to be. If the AI revolution unfolds gradually, there will be time for workers to adapt. Older professionals can finish out their careers, while younger ones can learn relevant skills or pivot to new fields. If the impact stays limited to certain sectors, it will be easier for workers to find opportunities in other parts of the economy.
But a broad and rapid shift will leave little time for adaptation and few places to take shelter. As Alex Imas put it, if the pace is slow, there’s time for employment to adjust and new roles to be created — it’s disruptive, but nothing humanity hasn’t seen before. Now, if it’s fast, truly unpredictable things could start to happen.
Public policy: the knot that still hasn’t been untied
If there’s one point of agreement among the economists who are revising their positions on Artificial Intelligence, it’s this: public policies are behind. Way behind. While private companies race to adopt, integrate, and scale AI across their operations, governments — in virtually every country — are still at the stage of understanding the problem, far from presenting concrete, actionable solutions. This gap isn’t new in the history of disruptive technologies, but the current pace makes this delay potentially more damaging than in previous eras.
Robert Seamans, an economist at New York University, was blunt: there’s enough discussion about the topic at this point for the country to start talking about what kinds of policies make sense in a world where how employment and careers work could change dramatically in the next two to five years.
A concrete example: the unemployment insurance system in the United States excludes many of the recent graduates who are likely to be the first ones hit by AI. Retraining programs are often slow and underfunded. These are tools that were designed for a different reality and need to be modernized to deal with the scenario that’s taking shape.
Anton Korinek, an economist at the University of Virginia, goes further. He argues that in the past, the social safety net was designed to help people weather transitional shocks — a period of temporary unemployment before finding a new position. But the AI shock could be more permanent. Korinek was one of the first economists to argue that AI could prove to be a uniquely transformative technology, and he continues to be a voice outside the consensus by considering more extreme scenarios, such as the possibility that AI could become better than humans at all tasks.
Many economists steer away from these discussions, something Korinek describes as emotionally understandable but practically a terrible idea. Part of the job of economists, he argues, is to worry about the biggest risks: what could cause disruptions and how we should prepare for them. In a move that says a lot about the current moment, Korinek is leaving the university at the end of the semester to work at Anthropic, one of the leading AI companies.
The discussion around public policies to mitigate the effects of AI on the labor market spans a range of fronts still being debated without much consensus. Professional retraining programs, education system reforms, creation of safety nets for workers displaced by automation, regulation of corporate AI use, and even bolder proposals like automation taxes or universal basic income are all on the table — but none of them have advanced consistently on a national scale, let alone globally.
What this means for people in the workforce right now
The message economists are sending isn’t one of panic — but it’s not one of comfort either. It’s a message to pay attention. The window of time to prepare exists, but it has an expiration date. Professionals across all fields, especially those with higher exposure to cognitive and repetitive tasks, have every reason to watch closely how Artificial Intelligence is being integrated into their industries. Not because the end is near, but because understanding this shift early is what separates those who will ride this wave from those who will be caught off guard.
In Brazil, the picture is even more delicate. The country has a labor market structure with widespread informality, limited social protection coverage for self-employed workers, and an educational foundation that hasn’t yet produced enough professionals to meet the digital demands that already exist today — let alone those coming tomorrow. This means the impact of technology-driven unemployment could be felt more intensely and more unevenly there than in economies with stronger qualification and social protection frameworks.
The good news — and it does exist — is that AI is also creating new demands. Professionals who know how to work with AI, who understand its limitations, who can supervise, adjust, and contextualize what these tools produce are becoming increasingly valued. The ability to combine human knowledge with the smart use of AI tools is, today, one of the most sought-after skills in the global labor market. That doesn’t cancel out the risks, but it opens a real path for those willing to move.
What economists are really asking for, at the end of the day, isn’t for people to fall apart in the face of technology — it’s for society, businesses, and governments to stop acting like this is a problem for the distant future. The signals are here. The economy is being reshaped in real time. And the difference between a well-managed transition and a social crisis of major proportions will largely come down to decisions that need to be made now — not after the unemployment numbers are already in the headlines. ⚡
