For the past two years, sensational announcements about artificial intelligence have fueled a recurring debate: Will AI lead to massive job losses? Some alarmist analyses are already predicting the rapid disappearance of skilled jobs, particularly in intellectual professions. However, when we examine the available data, the picture appears much more nuanced.
Between doomsday scenarios, the first visible changes in the labor market, and more optimistic long-term projections, one thing seems certain: AI is not causing an immediate collapse in employment, but it is accelerating a profound shift in the skills and occupations required.
Doomsday scenarios that fuel the debate
The recent release of a report by the U.S. firm Citrini Research has helped reignite concerns. The report outlines a scenario in which artificial intelligence systems would massively replace white-collar jobs within two years, causing a sharp drop in consumer spending and a contraction in U.S. gross domestic product of up to 70%1.
Although the authors present this scenario as a thought experiment rather than an economic forecast, it illustrates the concerns raised by the rapid advances in cognitive automation. Under this scenario, AI would replace a large portion of the intellectual tasks currently performed by skilled workers.
However, this type of projection is based on a bold assumption: that workers will be rapidly and widely replaced by autonomous systems. Yet the history of technological change generally shows that these processes are more gradual, with technologies transforming jobs far more than they immediately eliminating them.
Signs of change are already visible
While extreme scenarios remain debatable, some empirical data nevertheless show that artificial intelligence is already beginning to change the way the labor market operates.
A study by the Stanford Digital Economy Lab analyzed payroll data from the service provider ADP to track employment trends in the occupations most vulnerable to AI-driven automation. The results show a decline of approximately 13% in employment among recent graduates aged 22 to 25 in these professions since the end of 20222.
In certain fields, such as software development, the decline can reach nearly 20% for this age group. This trend is not the result of mass layoffs, but rather a shift in hiring practices: companies are hiring fewer entry-level employees when certain entry-level tasks can be automated.
At the same time, more experienced workers appear to have been relatively spared by this trend. In several occupations at risk from AI, employment among workers over the age of 30 has actually increased. This suggests that automation primarily affects routine or highly codified tasks, while activities that rely on experience, expertise, or human judgment remain more difficult to replace.
When AI replicates certain human decisions
Other research highlights the growing ability of AI systems to replicate certain professional decisions. For example, the Harvard Business School’s“Mimicking Finance” study showed that a machine learning model trained on financial data could replicate approximately 71% of the trading decisions made by traditional fund managers3.
This finding highlights the ability of algorithms to replicate certain decision-making processes based on historical data. However, it also reveals the current limitations of these systems: nearly 29% of decisions remain difficult to model. These choices often relate to more qualitative aspects of human work, such as intuition, the interpretation of contextual information, or strategic risk-taking.
In other words, AI can automate some analytical tasks, but it does not entirely replace the judgment developed through professional experience.
Overall positive long-term outlook
While certain professions are undergoing rapid transformation, overall employment projections remain relatively optimistic. The World Economic Forum’s *Future of Jobs 2025* report estimates that artificial intelligence and technological transformations could lead to the creation of approximately 170 million new jobs by 2030, compared with 92 million jobs that are likely to disappear4.
The overall net effect would therefore be positive, with approximately 78 million additional jobs worldwide.
These new opportunities include roles in data-related fields, AI systems engineering, cybersecurity, the energy transition, and services with a strong human element. Many companies are also planning large-scale training and reskilling programs to help their employees adapt to new technologies.
From this perspective, artificial intelligence appears to be more of a force for transforming the workplace than a driver of mass job destruction.
A transformation of work rather than a loss of jobs
When we look at all this data as a whole, one thing becomes clear: artificial intelligence is changing the nature of tasks within professions rather than completely replacing them.
Tasks based on explicit rules, the analysis of structured data, or repetitive procedures are the easiest to automate. In contrast, skills related to creativity, interpersonal relationships, complex decision-making, or the interpretation of new situations remain largely dependent on human capabilities.
This trend is reminiscent of previous technological revolutions. The computerization of the 1980s and the rise of the internet in the 2000s profoundly transformed many industries while creating new economic sectors.
In the case of AI, the main issue is therefore not so much the loss of jobs as the ability of individuals and organizations to adapt to this rapid transformation.
Ethical Issues: A Transition That Needs Support
Beyond economic considerations, the transformation of work driven by artificial intelligence also raises significant social and ethical issues. The main risk identified by many researchers is not necessarily massive job losses, but the emergence of new inequalities.
People who have access to training, digital skills, and technological tools could benefit greatly from the opportunities offered by AI. Conversely, those with fewer educational or economic resources may find it more difficult to adapt.
From this perspective, public policies, educational institutions, and businesses play a crucial role in supporting this transition. Investing in education, continuing education, and the dissemination of digital skills appears to be essential to ensuring that the benefits of artificial intelligence are shared by as many people as possible.
Training in AI skills: a key challenge
In this context, training emerges as one of the key drivers for supporting the transformation of the labor market. Skills related to artificial intelligence, data analysis, and the use of digital tools are gradually becoming cross-functional skills across many professions.
In fact, several studies show that workers with AI skills already enjoy significant advantages in the job market, particularly in terms of pay and employability.
Rather than leading to widespread job losses, AI therefore appears to be ushering in a new phase of transformation in the world of work, in which proficiency in technology is becoming a key factor in professional adaptation and development.
Learn more
The impact of artificial intelligence on employment is rarely understood on a job-by-job basis; rather, it is best understood through the concrete transformation of skills, tasks, and organizations. To broaden the analysis beyond the overall figures, we invite you to explore our section “AI & Professions”, which examines, profession by profession, how AI is reshaping roles, giving rise to new areas of expertise, and shifting value creation toward management, oversight, and decision-making activities.
References
1. Citrini Research. (2025). The Global Intelligence Crisis Scenario.
https://www.citrini.com
2. Stanford Digital Economy Lab. (2025). Canaries in the Coal Mine: AI and Early Career Employment.
https://digitaleconomy.stanford.edu
3. Harvard Business School. (2025). Mimicking Finance: Machine Learning and Investment Decisions.
https://www.hbs.edu
4. World Economic Forum. (2025). Future of Jobs Report.
https://www.weforum.org
