Emails can be drafted in seconds. Meeting contents are automatically organized, and response suggestions to inquiries appear on the screen. Yet, in the workplace, the cry of "we're short-staffed" is heard again today.

This scene does not necessarily signify a failure of AI. The ability to process parts of work faster and having the necessary personnel in the right place are separate issues.

On September 14, 2026, an article by dts Nachrichtenagentur published on the German financial information site Aktiencheck conveyed the views of Enzo Weber, an economist at the Institute for Employment Research (IAB). According to the article, he expressed to the German newspaper Rheinische Post that digitalization alone cannot eliminate labor shortages.

However, this does not deny the potential for improvement through AI. Weber acknowledges that AI can aid in productivity, business processes, and enhancing prosperity. He further pointed out that policies to nurture talent and expand employment opportunities cannot be replaced by technology alone.

By interpreting research and public comments starting from this statement, it becomes clear that the issue is not a binary choice between "AI or humans." It is about who uses the resources freed up by labor-saving measures and for what purpose.

The Distance Between "Reduced Workload" and "Eliminated Labor Shortage"

A single job comprises tasks of different natures. Creating documents, keeping records, listening to others' circumstances, checking the site, and handling exceptions. Even if AI can efficiently handle certain parts, the total number of people needed for the job doesn't change.

For instance, consider a job involving equipment inspection. If drafting reports becomes faster, the burden on the person in charge is reduced. However, the time taken to travel to the site, actually inspect the equipment, and verify the cause of anomalies may not be shortened by the same proportion. This is not a case study of a specific company but an example to separate tasks from professions.

Moreover, if the time saved is fragmented, there is a need to reorganize the work schedule. Several minutes of spare time created multiple times a day can be used for breaks or thorough checks, but it does not equate to having enough manpower to handle another site.

What becomes effective here is the perspective of not only "which professions will disappear" but also "which tasks will change."

A study published by the International Labour Organization (ILO) in 2025 estimated that about one in four workers globally is in an occupation that could be somewhat affected by generative AI. However, this is not a prediction that one in four will become unemployed. Since human involvement remains in the tasks that make up jobs, the study suggests that changes in jobs, rather than complete replacement, will likely be the focus.

The starting point of the discussion is to avoid jumping to the conclusion that "because there are tasks AI can handle, the person responsible for that job becomes unnecessary."

There is evidence of productivity improvement, but it is not omnipotent

Research shows that AI can enhance on-site capabilities.

The 2025 edition of the paper "Generative AI at Work" by Erik Brynjolfsson and others analyzed data from 5,172 customer support representatives. It was found that when they could utilize support from generative AI, productivity measured by the number of issues resolved per hour increased by an average of 15%.

The effect varied among individuals, with greater improvements seen among less experienced or lower-skilled representatives. On the other hand, the same level of effect was not observed among experienced and highly skilled representatives.

This result is suggestive when considering labor shortages. AI can be used not only to completely replace people but also to support the learning process of jobs, enabling more people to take on tasks.

However, the research subject was customer support at a single company, and it does not guarantee a 15% improvement across all workplaces. The authors also explain that it is not a study clarifying the impact on employment or wages across the entire economy.

When measuring the outcomes of AI, it is insufficient to only look at usage frequency or the speed of text generation. It is necessary to consider the total time required, including verification and correction, the number of issues resolved, and the quality of responses. This point is a proposal for applying research results to the workplace.

Even if drafts are made quickly, if corrections take time, the surplus capacity is small. Conversely, if appropriate advice is obtained in situations where newcomers stumble, it can help both the individual and their mentor. The question is not about introducing AI, but where it can be useful.

Efficiency improvements may lead to an increase in the work itself

One reason for the persistent labor shortage is changing demand.

Suppose a company can handle more work in the same amount of time thanks to AI. The company might use the extra time to reduce staff or shorten working hours. However, it could also choose to acquire new customers, accept previously declined requests, or start a different service.

If the cost of providing services decreases, it may become accessible to people who previously couldn't use it. If that happens, even if the workload per case is lighter, the number of cases handled may increase.

This is not an inevitable outcome, but it explains that the calculation "if productivity increases, the surplus workforce will increase accordingly" may not hold true for society as a whole.

The AI scenario analysis for Germany, published by IAB in November 2025, also separates the total amount of employment from its content. Under certain assumptions, if AI utilization progresses, approximately 1.6 million jobs will be affected by structural changes in the form of creation or disappearance over the next 15 years, while the total number of jobs will remain roughly the same compared to the baseline future path.

The figure of 1.6 million does not mean a net decrease of that number. It is also not an actual outcome but an estimate based on assumptions about AI introduction and other factors.

What can be inferred from this study is that even if the total number does not change significantly, the location and content of jobs can shift. Surplus personnel in one field and continued recruitment difficulties in another can occur simultaneously.

Surplus personnel cannot simply move to workplaces with shortages

Solving labor shortages through simple addition and subtraction of numbers is difficult.

For example, someone who has spare time due to efficiency improvements in clerical work cannot start handling equipment repairs the next day. New jobs require knowledge and experience, as well as considerations of work location, time, and personal preferences.

Therefore, if AI introduction and talent development are pursued separately, a gap may remain between departments that have achieved efficiency and those that need personnel.

An important consideration in this analysis is not only whether there are jobs to transfer to but also whether the process of transferring can be supported. Can time be secured for training? How will evaluations be handled until proficiency is achieved? Are arrangements in place to balance with childcare and family circumstances? Without these conditions, even if there are job openings and job seekers, they may not connect.

Weber's original article also highlighted the conditions under which people can demonstrate their abilities. He stated the need to better utilize career development opportunities for women and immigrants and to create an environment where older employees can continue working with prospects rather than increased burdens.

Translating this point into practice would mean considering training time, mentoring systems, and flexible work arrangements alongside the budget for AI implementation. It's too late to realize "there's no one who can use it" after introducing the technology.

Expectations and Questions Seen on Social Media—"What to Delegate and What Work to Retain"

What are the reactions on social media? Within the scope of publicly available information confirmed this time, direct comments on the article in question from September 14 were not sufficiently confirmed. Therefore, the following will address related comments on a LinkedIn post by Weber about a program dealing with AI and work, posted approximately two years ago. This is not a reaction to the current report but a past discussion on the same theme, summarized in Japanese.

Christian Mueller viewed AI positively, especially in expanding human potential in research and education. Nicole Koeck-Maier also expressed expectations that AI could take on tedious repetitive tasks, allowing humans to better utilize their creative power.

On the other hand, Antoinette Weibel questioned how work would change with AI. Would people only tend to machines, or could they engage in more meaningful and human-appropriate activities? She stated that the choice is influenced not by AI itself but by companies and the state responsible for necessary regulations.

These are individual opinions on limited posts and do not represent the overall public opinion or the ratio of pros and cons on social media. However, it shows that people's interest extends beyond "will jobs be taken away" to what kind of work will remain after AI is introduced.

Applying this point to the workplace reveals different uses for the same labor-saving measures. If the freed-up time is used for careful customer service or training, the quality of work can be improved. Conversely, if only the target number of processed cases is raised, the busyness of workers may remain unchanged. This is a consideration based on the posts.

Even if AI brings prosperity, it doesn't automatically increase time off

The discussion on labor shortages is also connected to the issue of working hours.

According to the original article, Weber is cautious about the view that if AI becomes widespread, people's working hours will automatically decrease. While productivity improvements can lead to prosperity, and there is a possibility of choosing to reduce working hours by accepting reduced income, he does not believe such changes will naturally occur for everyone.

He points out that as prosperity increases, expectations, living standards, and demands also change.

What needs to be considered here is the difference between the surplus capacity generated by technology and the mechanisms that determine its allocation. Even if AI saves an hour, whether that hour becomes a vacation or additional work is not determined by software alone.

For example, a company with increased productivity has multiple options: raising wages, lowering prices, enhancing services, increasing investment, or shortening working hours. What to prioritize involves management decisions and workplace agreements.

Therefore, even an evaluation of "successful AI implementation" requires an explanation of what the goals were. An increase in sales and a reduction in employee burden must be verified separately.

The question is not only "how many people can be reduced"

When bringing this discussion to Japanese workplaces, the German estimates cannot be directly applied due to differences in occupational structures and employment practices. However, the perspective of separating work efficiency and talent acquisition can be considered regardless of company size or country.

What this article proposes is to have three questions before and after AI implementation.

First, what is the real obstacle? Is it the time taken to create documents, the long wait for confirmation, or the lack of people who can handle the site? Increasing tools without identifying obstacles may not improve the overall workflow.

Second, whose capabilities are expanded? Will experienced workers be able to work even faster, will newcomers learn the job quickly, or will people who couldn't participate due to time constraints be able to take on tasks? The expected effects will change the necessary support.

Third, what will the freed-up time be used for? Compensating for recruitment difficulties, reducing overtime, improving customer service, or training the next generation of talent. Efficiency becomes a concrete result only after deciding how to use it.

AI can support measures against labor shortages. However, it does not take on the judgment of who to train, under what conditions to work, and how to allocate the gained surplus capacity.

Only by preparing both the speed of completing tasks on the screen and the conditions for people to continue working on-site can AI progress lead to workplace improvement.

Source URL

1. Aktiencheck/dts Nachrichtenagentur "Ökonom: KI löst Fachkräftemangel nicht" (September 14, 2026). Confirmation of Weber's statements on labor shortages, working hours, and employment opportunities for women, immigrants, and older employees.
   https://www.aktiencheck.de/news/Artikel-Oekonom_KI_loest_Fachkraeftemangel_nicht-20090389

2. ILO "Generative AI and Jobs: A Refined Global Index of Occupational Exposure" (May 20, 2025). The basis for the estimate that about one in four workers globally is in an occupation that could be affected by generative AI, and the distinction between the disappearance of entire occupations and changes in tasks.
   https://www.ilo.org/publications/generative-ai-and-jobs-refined-global-index-occupational-exposure

3. Erik Brynjolfsson, Danielle Li, Lindsey Raymond "Generative AI at Work" (The Quarterly Journal of Economics, 2025 edition). Data from 5,172 customer support representatives, an average 15% productivity improvement, differences in effects due to experience, and the limitation that results from a single company cannot be generalized to employment across the entire economy. Refer to the authors' publicly available PDF.
   https://danielle.li/assets/docs/GenerativeAIatWork.pdf

4. IAB "Künstliche Intelligenz könnte das BIP-Wachstum in den nächsten 15 Jahren um jährlich 0,8 Prozentpunkte erhöhen" (November 19, 2025). The basis for the conditional future analysis that approximately 1.6 million jobs will be affected by creation or disappearance, while the total number of jobs will remain roughly stable compared to the baseline scenario.
   https://iab.de/presseinfo/kuenstliche-intelligenz-koennte-das-bip-wachstum-in-den-naechsten-15-jahren-um-jaehrlich-08-prozentpunkte-erhoehen/

5. Enzo Weber's LinkedIn post "Nimmt uns KI die Arbeit weg?". Confirmation of the introduction of a program dealing with AI and work, and public comments by Christian Mueller, Nicole Koeck-Maier, and Antoinette Weibel. The page displays "2 years ago." Used as a past discussion on related themes, not a direct reaction to the original article from September 14, 2026.
   https://de.linkedin.com/posts/enzo-weber_ki-arbeit-k%C3%BCnstlicheintelligenz-activity-7208388393196515328-VzFO