AI and Labor / Reading the European Commission's 'The future employment impact of artificial intelligence and emerging digital technologies in Europe' / Thinking about the fragmentation of labor and the reconstruction of law from the perspective of 2026

0 Introduction

In November 2025, the European Commission's Social Situation Monitor published a research note that cannot be overlooked when forecasting future AI legal policy. It is titled 'The future employment impact of artificial intelligence and emerging digital technologies in Europe' by Fabien Petit (University of Barcelona, UCL, CEPEO).
Now that the social implementation of generative AI has completed its first cycle and AI laws are in full operation in Europe, our questions have changed. It is no longer the vague question of whether AI will replace humans. The focus has shifted to which segments, which tasks, and at what speed things are changing. Petit's report uses the TechXposure approach, which connects patent data and job descriptions through natural language processing, to precisely analyze the impact that 40 emerging digital technologies have had on employment over the past decade or more.
In this article, I will examine the structural challenges facing labor law in the AI era while unraveling several facts revealed by this report. It is not just about the quantity of employment. I will delve into the quality of labor, fragmentation, and how to redefine what the law should protect.

1 The Location of the Problem

The debate surrounding AI and employment has repeatedly swung like a pendulum between pessimism and optimism.
Pessimists sound the alarm that AI will surpass humans in cognitive ability, causing widespread unemployment, including among white-collar workers. Optimists argue that technological innovation will increase productivity, creating new demand and job types, thereby maintaining and expanding employment in the long term.
The conclusion reached by Petit's report, from a macro perspective, supports the latter. Regions with higher exposure to AI and robotics showed an upward trend in the employment-to-population ratio, confirming a net increase in employment. The productivity effect outweighs the displacement effect.
However, from a legal perspective, what we should focus on is the asymmetry hidden behind macro averages. Even if employment is increasing overall, if workers with specific attributes are being structurally excluded, the law must intervene. The report highlights the structure of that exclusion.
What is shocking is the impact by age group. Employment increases are seen among the young (16-24) and the elderly (45-64). On the other hand, a negative impact on employment was observed in the core working-age group (25-44), who should be in their prime. AI is hitting the tasks of mid-level practitioners in the middle, rather than unskilled labor or skilled management. This hollowing out of the middle is the greatest risk facing the labor market in the AI era and will become the new main battlefield for labor law.

2 The Problem of Fragmentation and Identifying Causality

Why does the core working-age group get the short end of the stick? To understand this, it is essential to view labor not as a block of duties, but as a collection of tasks.
AI does not replace humans themselves. It replaces routine and predictable tasks within the work humans perform, pinpointing them. The 25-44 age group is the segment that carries the most routine cognitive tasks, such as document creation, data analysis, schedule adjustment, and primary decision-making, which form the core of practical work within an organization.
Task replacement does not immediately lead to dismissal. That is why the problem is difficult to see. Herein lies the problem of fragmentation.
Previous technological innovations, such as industrial robots, were visible replacements that took away the entire jobs of factory workers. The erosion by generative AI is different. It proceeds by gradually stripping away parts of a job. Workers witness their work being replaced by AI bit by bit, and the remaining tasks transform into checking AI output or handling complaints that AI cannot deal with. It is either low value-added or high stress, or both.
What becomes a legal issue is the difficulty of identifying causality in this process. When a worker complains that their work has become harder, their wages have not increased, or they have been transferred to an unwanted department, it is difficult to determine whether it is a structural change due to AI introduction, economic fluctuations, or an individual's ability. If an employer claims that your main tasks have disappeared due to AI and proceeds with demotion or redundancy, how should a court determine whether that task disappearance is true or whether the change justifies dismissal? Detailed task analysis data, such as that used by Petit, may emerge in the future as evidence for fact-finding in litigation.

3 Algorithmic Management and the Jurisprudence of Transition

The spread of emerging digital technologies shown in the report forces a transformation not only in the quantity of employment but also in its quality, that is, the nature of command and control.
This is the infiltration of algorithmic management. If AI performs task allocation, progress monitoring, and evaluation decisions, workers will be managed by an invisible boss.
In Europe, through the Platform Work Directive and the AI Act, efforts are underway to ensure the transparency of algorithmic decisions and mandate human oversight. However, this problem is not limited to platform work. It cannot be overlooked in general employment relationships either.
For example, if AI predicts that this employee's performance will decline in the future and, based on that, denies them educational training opportunities or removes them from a promotion list, is that legally permissible? It is profiling and statistical discrimination.
Labor law has traditionally regulated arbitrary evaluations by human employers. Because algorithmic evaluation is disguised as objective, discrimination and bias are easily preserved.
What is required here is the establishment of two legal rights.
First, the realization of the right to request an explanation. The right to know what parameters were used to determine one's treatment or evaluation is at the core of due process in the AI era.
Second, the guarantee of the right to career transition. As Petit's analysis shows, for segments whose skills become obsolete due to technological change, the right of access to prior reskilling carries more weight than subsequent monetary compensation. Whether the duty to cooperate in maintaining and improving continuous employability can be read into the labor contract as a duty corresponding to the employer's right of command and control will be the focus of interpretation.

4 Implications for Japanese Law

Turning our eyes to Japan, we see a different, but more serious, distortion than in Europe.
Japan is experiencing a chronic labor shortage due to a declining birthrate and aging population, and labor-saving through AI is welcomed. The net increase in employment that Petit mentions would likely be received positively in Japan as a mitigation of labor shortages.
However, there is a risk that the rigidity of the Japanese labor market will make this change cruel.
Japan's dismissal regulations are strict, and the employment of regular employees is strongly protected. At first glance, this looks like a breakwater against the risk of unemployment due to AI. However, precisely because dismissal is difficult, companies have an incentive to either keep surplus personnel on the payroll after AI introduction or force them to resign through legally gray means.
Because Japan's membership-based employment does not limit job duties, employers have a strong right to order transfers. An order saying, 'Your job has been replaced by AI, so do this job from tomorrow,' is easier to implement than in the West.
What is concerning is a situation where workers are exhausted by repeated transfers to jobs for which they are not suited, without being given sufficient opportunities for reskilling. Or, it could be a return of the 'employment ice age' where the employment of the middle-aged and elderly is protected while the hiring of the young is suppressed by AI, widening the intergenerational gap.
What is required of Japanese law is not a simple debate about relaxing dismissal regulations. It is to shift the center of gravity of policy from maintaining employment to supporting mobility. This includes expanding subsidies for labor mobility support, legalizing educational training leave, and activating the involvement of labor unions and worker representatives in algorithmic evaluation and placement.

5 In Conclusion

Petit's report tells us that the stage of fearing technological unemployment is over. What is depicted there is not a dystopia where the world is destroyed by AI. It is the reality of a fragmented labor market where winners and losers are selected quietly, but cruelly.
AI makes labor efficient and creates wealth. There is no guarantee that this wealth will be distributed fairly. If adjustment is left to the market, specific age groups and skill levels will be sacrificed. The data shows this clearly.
How can the law correct the inequality that technology accelerates? The mission assigned to labor law is not to stop AI. It is to design an institutional rectifier that converts the waves of change brought about by AI into energy that protects the dignity and lives of workers.
The question posed by the European Commission's research note is not a distant European story. Japan in 2026, where a declining birthrate and aging population and the spread of AI are progressing simultaneously, may be the place that needs to face this difficult problem most seriously.

6 Summary of AI and Labor

(Magazine) 'Random Thoughts on AI and Law'

※ Please refer to the following for the table of contents

note General Terms of Service Article 3, Paragraph 2, First Sentence
3.2 The copyright of digital content created by the creator belongs to the creator.