What the Harvard study revealed about AI and jobs
The research conducted by professor Suraj Srinivasan from Harvard Business School analyzed millions of job postings across the United States between 2019 and March 2025, making it one of the most comprehensive studies ever done on the real impact of generative Artificial Intelligence on the job market. Rather than relying on theoretical projections or hypothetical scenarios, the research team dug into concrete data from recruiting platforms to understand what actually happened with job availability after tools like ChatGPT hit the scene in November 2022. The results painted a picture that is more nuanced and, honestly, more interesting than the simple doom-and-gloom narrative we tend to see floating around.
The academic paper, titled Displacement or Complementarity? The Labor Market Impact of Generative AI, was developed in collaboration with Wilbur Xinyuan Chen from the Hong Kong University of Science and Technology and Saleh Zakerinia from Ohio State University. The researchers used a massive dataset covering virtually every job opening in the United States during the period analyzed, giving the study a breadth that is hard to find in similar work.
The study divided roles into two broad categories. On one side were the so-called positions with high automation exposure — jobs made up mostly of repetitive, standardized, and predictable tasks like data entry, routine customer service, and document processing. These positions saw a 13% drop in job postings after ChatGPT launched, confirming what many people already suspected. Generative Artificial Intelligence is extremely efficient when the work follows well-defined patterns, and it makes sense that companies would start automating these activities over time. But stopping the analysis at this point would mean telling only half the story.
On the other side were roles with high augmentation potential — analytical, technical, and creative positions where AI functions as a support tool rather than a replacement. And here is where the data gets surprising: demand for these professionals grew by 20% over the same period. We are talking about positions that combine automatable tasks with others that still require direct human involvement, such as critical thinking, decision-making, and the ability to interpret complex contexts. The study itself cites specific examples: microbiologists, financial analysts, and clinical neuropsychologists are among the professions with the highest augmentation potential. This shows that job creation linked to AI is not just a futuristic promise — it is something already happening.
How the research classified more than 900 occupations
One of the most interesting aspects of the methodology is that the researchers used OpenAI’s ChatGPT itself to categorize more than 19,000 professional tasks spread across over 900 different occupations. Each task was evaluated based on its potential to be automated by generative AI. From that classification, the team built an augmentation score based on the proportion of exposed and non-exposed tasks within each occupation.
This method allowed them to go beyond vague generalizations like AI is going to kill all the jobs and map out with granularity which specific roles are being most affected and in what ways. The results revealed that the biggest reductions in job postings occurred in the finance and technology sectors — precisely the fields that deal with large volumes of structured data and processes ripe for standardization. At the same time, within those very same sectors, new demand emerged for professionals capable of working alongside AI tools, reinforcing the dual nature of this transformation.
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.
Another noteworthy finding is that the number of skills required for positions with high automation exposure is shrinking. The researchers identified a 7% reduction in the competencies listed in those job postings, along with fewer emerging skills showing up in those occupations. In contrast, in roles with high augmentation potential, AI-related skills — like prompt writing and the use of artificial intelligence tools — are becoming increasingly common in job descriptions. As workflows transform with the new technology, entirely new competencies are also emerging.
Augmentation: when AI works with you, not against you
The concept of augmentation is central to understanding this transformation and deserves special attention. Unlike pure automation, which replaces human labor with a machine, augmentation operates on a collaborative logic. The idea is that Artificial Intelligence handles the more operational and repetitive parts of a role, freeing up the professional to focus on what truly requires human capability: creativity, empathy, ethical judgment, and strategic vision.
As professor Srinivasan himself explains, in the financial sector, investment managers and analysts use AI-powered tools to process and evaluate market data at unprecedented speed, but at the end of the day, judgment and decision-making remain crucial and essentially human. AI speeds up the process, improves the quality of inputs, and reduces operational errors, but the final interpretation — the one that accounts for contextual nuances, trust-based client relationships, and risk perception — stays in the hands of the professional.
This augmentation model is already being adopted by companies of different sizes and across various industries. In healthcare, doctors use AI systems to analyze imaging exams more quickly and accurately, but the clinical decision remains human. In the legal sector, natural language processing tools help lawyers review thousands of pages of contracts in minutes, allowing them to focus their energy on crafting more creative legal strategies. In marketing, professionals use generative models to create campaign drafts and analyze audience data, but the curation of tone, messaging, and brand positioning remains a responsibility that depends on human sensitivity and experience.
The most important takeaway here is that augmentation changes the very definition of what it means to be competent in a given profession. Mastering the traditional technical skills of your field is no longer enough. Now, you also need to know how to use AI tools to expand your output and make better decisions. This creates a new tier of productivity that benefits both professionals and organizations, and it explains why job openings in these high-augmentation-potential areas are growing so significantly.
What Srinivasan recommends for companies
The study does not stop at presenting numbers. Professor Srinivasan offers direct recommendations for organizations looking to navigate this transition wisely. How companies integrate generative AI technologies is decisive in determining whether the outcome will be job loss or job creation. And two fronts stand out:
- Invest in reskilling programs to retrain workers toward AI-augmented roles. According to Srinivasan, retraining is essential for occupations where generative AI is reducing the diversity of required skills. In automation-prone roles, workers may face displacement unless they develop non-automatable competencies, such as judgment and interpersonal communication skills.
- Promote ongoing training in generative AI so that professionals can take full advantage of the new tools. In augmentation-prone occupations, generative AI is expanding skill requirements, increasing demand for AI literacy, human-machine collaboration, and domain-specific artificial intelligence applications.
In Srinivasan’s view, companies should see generative AI as an augmentation tool, not merely as a cost-cutting measure. Workforce training programs need to be aligned accordingly, to support both career transitions and ever-evolving skill demands. This perspective completely changes how leadership should think about investing in people. It is not about spending less on teams because AI handles part of the work — it is about investing differently to get the most out of the combination of human talent and computational power.
Reskilling: the most important skill of the moment
If augmentation defines how the relationship between humans and AI will work in practice, reskilling is the path to getting there. And we are not talking about something optional or nice-to-have — we are talking about an urgent necessity. The Harvard study reinforces that professionals whose roles have been most impacted by automation are not necessarily doomed to unemployment, but they need to develop new competencies to transition into positions where AI complements rather than replaces them. This includes skills like data analysis, computational thinking, strategic communication, and, of course, the ability to work efficiently with Artificial Intelligence tools.
Major corporations like Amazon, Google, and Microsoft have already expanded their AI and data analytics training programs, many of them free and accessible online. In Brazil, public and private initiatives are also gaining momentum, with platforms offering reskilling courses aimed at professionals in fields like administration, logistics, and customer service — precisely the segments most exposed to automation. The challenge, however, is scale. Millions of workers need to be retrained within a relatively short timeframe, and that requires coordinated investment, smart public policies, and a corporate culture that values continuous learning.
For those in the middle of this transition, the most important thing to understand is that reskilling does not necessarily mean abandoning your field and learning to code from scratch. In many cases, it is about incorporating new tools into the skill set you already have. An HR professional who learns to use AI for resume screening and organizational climate analysis, for example, becomes more strategic without needing to become a machine learning engineer. A designer who masters generative image tools can deliver more projects at higher quality without losing the authorial identity of their work. The key is identifying which competencies in your current field can be enhanced by technology and pursuing training in that direction. 🚀
Limitations of the study and what we still don’t know
The researchers themselves are careful to point out that the study focuses on the short-term impact of generative AI on the U.S. job market. This means that the effects in other regions of the world, as well as long-term impacts, remain uncertain as technology adoption advances on a global scale. This is an important caveat, because the pace of AI adoption varies enormously across countries, industries, and company sizes.
Additionally, the American job market has specific characteristics — such as greater flexibility in hiring and firing — that may not translate to economies with different labor laws, like Brazil. This does not invalidate the research findings, but it calls for caution when generalizing the results to other contexts. What we can say with confidence is that the trends identified by the study — automation of repetitive tasks and growing demand for human-AI collaboration — are global forces that will impact job markets worldwide to varying degrees.
What to expect from the job market going forward
The Harvard data make it clear that the narrative of Artificial Intelligence simply destroying jobs on a massive scale is, at the very least, incomplete. As Srinivasan himself summarizes, rather than just eliminating jobs, generative AI creates new demand in augmentation-prone roles, suggesting that human-AI collaboration is a central engine of labor market transformation. What is happening is a deep reorganization, where some roles lose relevance while others emerge or expand at an impressive pace.
The net outcome of this equation is still being calculated, but the early signs point to a scenario where job creation in areas of high cognitive complexity outpaces losses in routine activities. This does not minimize the impact for those working in roles directly threatened by automation, and it is essential that this portion of the workforce receives adequate attention and support. But the overall trend suggests that AI could, paradoxically, be one of the biggest job-creation forces of the next decade.
The key takeaway for companies is understanding that investing in augmentation and reskilling is not just a matter of social responsibility — it is a competitiveness strategy. Organizations that train their teams to work with AI can retain talent, boost productivity, and position themselves better in increasingly dynamic markets. For professionals, the message is equally clear: technology moves fast, but the human ability to learn, adapt, and create value in new contexts remains the most valuable asset anyone can have. The future of work does not belong to robots, nor does it belong solely to humans. It belongs to those who know how to combine the best of both worlds. 💡
