AI transforms work, but does not fundamentally eliminate it
Uma das maiores transformações do mercado de trabalho em gerações está em curso, impulsionada não por guerras ou crises financeiras, mas por algoritmos capazes de aprender. O Fórum Econômico Mundial projeta que a inteligência artificial eliminará 92 milhões de empregos até 2030, ao mesmo tempo em que poderá criar 170 milhões de novos — um saldo positivo no papel, mas profundamente desigual na prática. O que está em jogo não é apenas a sobrevivência de certas ocupações, mas a capacidade das sociedades de preparar seus trabalhadores para um mundo onde o conhecimento técnico, antes um escudo, tornou-se o primeiro alvo.
- Gigantes como Amazon, Microsoft e Duolingo já estão reduzindo equipes ou substituindo funções por algoritmos, sinalizando que a disrupção não é futura — ela já começou.
- Pela primeira vez em ondas tecnológicas anteriores, são os trabalhadores do conhecimento — programadores, contadores, analistas — os mais vulneráveis, invertendo a lógica histórica que penalizava os menos qualificados.
- Economias desenvolvidas enfrentam o maior impacto, com 60% da força de trabalho afetada, enquanto países de baixa renda, menos expostos, também colherão menos dos ganhos de produtividade prometidos pela IA.
- Pesquisadores como Enzo Weber e economistas de Harvard argumentam que a automação transforma o trabalho mais do que o elimina, desde que empresas e trabalhadores adotem ativamente as novas ferramentas.
- O verdadeiro divisor de águas não será a tecnologia em si, mas a disposição humana de se adaptar: sem treinamento e integração efetiva, os ganhos prometidos podem nunca se materializar.
O presidente-executivo da Amazon, Andy Jassy, foi direto ao ponto: a inteligência artificial vai remodelar o emprego em muitos setores. Daniela Amodei, cofundadora da Anthropic, foi ainda mais específica, prevendo que a IA pode eliminar metade dos cargos de nível inicial no trabalho de escritório em até cinco anos. O padrão já é visível: desde 2022, empresas americanas de capital aberto reduziram seus quadros presenciais em 3,5%, e um quinto das companhias do S&P 500 encolheu. Shopify exige que equipes provem que a IA não pode fazer o trabalho antes de contratar alguém novo.
O Fórum Econômico Mundial projeta que a IA eliminará 92 milhões de empregos globalmente até 2030, mas criará 170 milhões de novos. O FMI estima que economias desenvolvidas serão as mais afetadas, com 60% da força de trabalho impactada. A OCDE calcula que um quarto de todos os empregos no mundo pode se tornar obsoleto. Desta vez, os mais vulneráveis não são os trabalhadores manuais de baixa qualificação, mas os profissionais de escritório — programadores, contadores, analistas de dados — funções onde os algoritmos já igualam ou superam o desempenho humano.
Nem todos, porém, enxergam catástrofe. O pesquisador alemão Enzo Weber argumenta que a IA transforma o trabalho, mas não o elimina fundamentalmente — ela ajuda humanos a desenvolver novas tarefas e executá-las melhor. Um estudo com participação dos economistas de Harvard David Deming e Lawrence Summers reforça essa visão: automatizar tarefas pode até gerar ganhos de emprego em alguns setores, pois trabalhadores mais produtivos geram produção adicional que compensa o que foi automatizado.
O desafio real está na integração. Se trabalhadores e empresas não adotarem plenamente essas ferramentas, os ganhos de produtividade prometidos podem nunca chegar. Weber é enfático: treinamento ativo e desenvolvimento de pessoas não são opcionais — são a condição para que a IA seja uma oportunidade e não apenas uma ameaça. O desfecho depende menos do que a tecnologia é capaz de fazer e mais do que os humanos escolherem fazer com ela.
Amazon's chief executive Andy Jassy announced recently that the company would cut more jobs as artificial intelligence takes over tasks once performed by humans. He was blunt about it: the technology will reshape employment across many sectors. He is not alone in this assessment. At the AI startup Anthropic, president and cofounder Daniela Amodei said in May that within one to five years, artificial intelligence could wipe out half of all entry-level positions in white-collar work. The pattern is visible across the economy. Since 2022, public companies in the United States have reduced their in-office workforce by 3.5 percent. One-fifth of the companies in the S&P 500 have shrunk their headcount. Microsoft, Hewlett Packard, and Procter & Gamble have all announced thousands of layoffs. Shopify now requires teams seeking to hire new people to prove that AI cannot do the job. Duolingo is gradually replacing external workers with algorithms.
The World Economic Forum released a forecast at the start of this year that captures the scale of the disruption ahead. Artificial intelligence could eliminate 92 million jobs globally by 2030. But the same technology, the forum predicts, could create 170 million new ones. The International Monetary Fund conducted its own analysis in 2024 and found that developed economies would feel the heaviest blow, with 60 percent of their workforce affected—half negatively, half positively. Emerging economies would see 40 percent of jobs touched by AI. In low-income countries, the figure drops to 26 percent. Yet there is a catch: those countries that face less disruption also stand to gain less from the productivity gains AI promises.
The Organization for Economic Cooperation and Development estimates that one-quarter of all jobs worldwide could become obsolete because of artificial intelligence. But which jobs, exactly? Research from the Pew Research Center identified the most vulnerable occupations: those involving data collection and analysis, such as web programming, technical writing, accounting, and data entry. The jobs most likely to survive are those demanding intense manual labor and resistance to automation—construction workers, childcare providers, firefighters. This marks a shift from previous waves of technological disruption. In the past, the workers hit hardest were those with less education and manual laborers, like factory workers replaced by robots. This time, the threat reaches higher up the ladder, targeting office workers and knowledge workers in roles where algorithms now match or exceed human performance.
But not everyone sees catastrophe ahead. Pope Francis warned of the risks AI poses to human subsistence and dignity. Yet labor market experts offer a different view. Enzo Weber, a researcher at Germany's Institute for Employment Research, argues that these fears are overblown. He contends that technological progress opens broad economic possibilities and may actually help workers rather than trigger mass unemployment. "AI primarily transforms work, but does not fundamentally eliminate it," Weber said. In many cases, he argues, the technology assists humans in developing new tasks and executing them better, rather than replacing them outright.
A study published in January supports this perspective. Its authors, including Harvard economists David Deming, Christopher Ong, and Lawrence Summers, argue that automating tasks at work does not necessarily reduce employment. It can even generate job gains in some sectors. When a laborious task becomes automated, workers become more productive, and the additional output they generate can offset the portion of their work now done by machine. The economists acknowledge that AI's impact will likely be "sweeping and lasting," but they note that history teaches us such disruptions unfold over many decades, not overnight.
The real challenge lies in how AI gets integrated into workplaces and whether workers embrace it or resist it out of fear for their jobs. If employees and companies fail to adopt these tools fully, the promised productivity gains may never materialize. Weber urges both workers and employers to adapt to the shifting technological landscape and seize the opportunities it presents. He calls AI a "game changer" that offers possibilities—but only if people reach out and grab them. "Active development and training of employees is essential," he said. "Not just to keep pace, but to advance as much as possible." The outcome depends less on what the technology can do and more on what humans choose to do with it.
Notable Quotes
AI primarily transforms work, but does not fundamentally eliminate it. In many cases, it helps humans develop new tasks and execute them better, rather than replacing them.— Enzo Weber, Institute for Employment Research, Germany
Automating a laborious task can make workers so much more productive that the additional output compensates for the portion of work now done by machine.— David Deming, Christopher Ong, and Lawrence Summers, Harvard University
The Hearth Conversation Another angle on the story
When Amazon and Anthropic say they're cutting jobs because of AI, are they being honest about the cause, or is AI just a convenient excuse for decisions they'd make anyway?
That's the right question to ask. The timing is suspicious—these announcements cluster around the moment AI became genuinely capable. But it's also true that companies have always looked for ways to reduce labor costs. What's different now is that AI actually works well enough to do certain jobs. So it's probably both: AI is real, and it's also a useful narrative for restructuring.
The World Economic Forum says 92 million jobs disappear but 170 million appear. That sounds almost optimistic. Why should anyone believe those numbers?
You shouldn't believe them with certainty. They're forecasts, not prophecies. But the logic behind them is sound: new technology creates new categories of work that didn't exist before. The internet destroyed some jobs and created others—web designers, social media managers, cloud engineers. The question isn't whether new jobs appear. It's whether they appear fast enough, in the right places, and whether displaced workers can actually transition into them.
So a 55-year-old accountant whose job gets automated—what happens to them?
That's where the optimism breaks down. The studies assume workers can retrain, move, and find new work. But retraining takes time and money. Moving takes resources. And a 55-year-old competing for a new entry-level role in a different field faces real barriers. The aggregate numbers might work out. Individual lives might not.
Why do construction workers and childcare providers seem safer from automation?
Because those jobs require physical presence, judgment in unpredictable situations, and human touch in ways algorithms struggle with. A robot can do repetitive factory work. It's much harder to build a robot that reads a child's emotional state or navigates a chaotic construction site. That doesn't mean they're safe forever—just that the timeline is longer.
Enzo Weber says AI transforms work rather than eliminating it. Is he being naive?
He might be. But he's also drawing on historical precedent. The printing press, electricity, computers—all were supposed to destroy work. They didn't, at least not permanently. They shifted it. The difference this time is speed and breadth. AI could affect white-collar work faster than previous technologies affected blue-collar work. That compression matters.
What would actually need to happen for the optimistic scenario to come true?
Massive, sustained investment in worker retraining. Policy that doesn't let displaced workers fall through cracks. Companies willing to hire and train people for new roles rather than just cutting costs. And workers willing to learn new skills instead of retreating into fear. None of that is guaranteed.