Two visions clash in conversations about the future of work. On the one hand, growth thanks to technology. On the other hand, there is shrinking space for employees. The “Canaries in the Mine” study published last year confirms that concerns about the future of work are justified. In the 22-26 age group, in digital professions most exposed to AI – such as call centers and programming – employment fell by 13%. Now, in Davos, prof. Erik Brynjolfsson from Stanford University updated the data: within a few months, the employment gap had widened to 16 percent. The specter of “jobless growth” is becoming more and more real.
What can be done to ensure that productivity benefits are not achieved at the expense of eliminating jobs? The answer from Davos is that the key lies not in what AI can do, but in how companies implement it.
Why can't AI act on its own? The role of humans in automation
Experts from Davos point out that the very nature of the new technology requires the presence of humans. Ravi Kumar, CEO of Cognizant, pointed out the fundamental difference between artificial intelligence and traditional software. In the past, systems worked according to a rigid logic and always gave the same result. AI works differently – it provides answers with a certain level of accuracy.
In practice, this means that the system can suggest a solution with “87 percent confidence.” In medicine, industry or finance, it is not enough for a machine to operate independently. Therefore, companies must switch to a model that Kumar calls “macro-delegation and micro-control”: a person delegates tasks to the system, but checks the result, context and takes responsibility for the final decision.
Jonas Prising, CEO of ManpowerGroup, used an apt aviation metaphor here: although statistically, pilots make more mistakes than the autopilot, as passengers we will not accept an airplane without a human in the cockpit. We need human judgment in critical situations.
How to implement AI without losing employee knowledge
If humans are essential, why do companies pursue automation? Prof. Brynjolfsson called this the “Turing trap”. It's an obsession with creating machines that perfectly imitate humans, rather than ones that expand their capabilities. This is not only a social mistake, but also a business one.
For AI to be truly useful, it must understand the context of the company. Ravi Kumar calls it “tribal knowledge” or “the dynamics of everyday work”. This is not in the model training data. Only people have this knowledge: they know the procedures, risks and informal rules of operation of the organization.
That's what it's all about contextual engineering: on teaching artificial intelligence systems the realities of a specific company and involving employees in the process of their design.
Read also: Christine Lagarde in Davos spoke about AI, wealth and international cooperation
AI implementation and employees. Technology cannot be imposed from above
During the debates on the future of work in Davos, it was repeatedly emphasized that artificial intelligence should not be implemented solely by the decision of management boards and technology providers. It is crucial involving employees in this process from the very beginning — before the systems reach production halls, offices or driver's cabins.
Liz Shuler, president of the American federation of trade unions AFL-CIO, cited the example of cooperation with Microsoft, in which bus drivers work side by side with programmers in research laboratories. The goal is to check whether AI actually helps driving or whether it becomes a tool for micromanagement and punishment for every mistake.
— We are not against technology. We want to make sure that technology doesn't roll over us. In every industrial revolution, it was workers who helped “tame” the machines, Shuler said.
Read also: Burnout is a matter of time. Corporations are squeezing managers
Employee rights in the age of AI. The role of trade unions and data protection
There is also a new battlefield for employee protection how the data is used. Luc Triangle, secretary general of the International Trade Union Confederation, emphasized that MrEmployees have the right to view and decide what data their employers use and for what purpose. This is a fundamental protection of another person's privacy and dignity.
“None of us would want an employer to check how much time we spend in the toilet,” Triangle concluded bluntly, pointing to the risk of total surveillance in the blind pursuit of productivity.
In turn, Liz Shuler drew attention to threat to professional autonomyemphasizing the role of collective agreements in setting the rules for data use. As a warning, she cited the WNBA basketball league, where AI begins to evaluate every decision made by players on the court. This risks denying athletes — and workers in general — the right to professional judgment and intuition.
Who will lose their job because of AI? Threatened industries and the speed of market changes
Who will lose their job because of AI? Intuition suggests that those who sit in front of computers.
— We told our children that their future was in science and technology. STEM was supposed to be a guarantee of employment stability. Meanwhile, this is where the fastest changes are taking place, noted Ajay S. Banga, president of the World Bank.
Ravi Kumar, however, does not share these concerns. At Davos he emphasized the key distinction between exposure on technology a speed of change (velocity).
Paradoxically, the IT sector – although it has huge exposure to AI – has already seen the worst. The change happened quickly there, developers learned to use new tools, and the industry “reset” its expectations.
In his opinion, a real shock may await physical sectors such as construction, health care, industry and transport. Today, their exposure to AI seems low, but when the technology finally gets there, the speed of change will be shocking.
– It is there – and not in air-conditioned offices – that the second, much more violent wave of transformation may come, which will find employees and companies completely unprepared – he concluded.
Read also: Great Britain is betting on artificial intelligence. The number of job offers is growing
In turn, the President of Singapore, Tharman Shanmugaratnam, called for a reevaluation of the global social ladder. In his opinion we have to end the hierarchy that places academic education and intellectual work above others. Artificial intelligence equalizes these opportunities.
Interestingly, the data shows that employees with lower qualifications benefit the most from working with AI (productivity increase by approximately 35%), because the technology allows them to jump over entry barriers faster and gain expert knowledge.
Read also: Where is it easiest to find a job? List of professions 2026
The future of the labor market. The role of policy and employee reskilling
Although experts in Davos agree that in the long term the technology will not cause permanent unemployment, the transition period may be painful and requires state intervention.
Economists Laura Tyson and Liz Shuler warn that previous transformations led to labor market polarization and a decline in the share of wages in GDP. This time, the benefits of new productivity must be shared more fairly.
Valdis Dombrovskis, the EU Commissioner for Economy and Productivity, pointed out that for an aging Europe, artificial intelligence can be a tool for maintaining economic growth with a shrinking number of employees. There is a condition though mass retraining and investment in competences before technology starts to displace people from the labor market.
At stake is no longer just the number of jobs, but their quality. The question is whether working with AI will be safe and whether one full-time job will be enough to support a family with dignity.
Responsibility lies with both employers and those in power. The global labor market is changing on an unprecedented scale. The vision of a world without work, although tempting in its utopian form, is (so far) unrealistic. Regulations will play a key role to protect us against fully automated economic growth.
