A door that opens less often
A call center can lose staff without announcing a major layoff: it only needs to hire fewer people than leave. That is the change documented by Caelan Wilkie-Rogers, an analyst at Revelio Labs, in research published on October 6, 2026. Observed global employment is 5.3% below its December 2023 peak. In the twelve middle-income countries analyzed, entries decline faster than exits. For someone seeking a first job, the consequence may be a door that opens less often.
The innovation making this question urgent is automated assistance: systems capable of answering and handling requests previously assigned to operators. The research observes the labor market, rather than each individual deployment. It does not prove that every missing job has been replaced by a chatbot. It does, however, reveal a specific problem: while other office occupations grow, people leaving call centers rarely reach the better-paid ones.
Movement is the most important measure
Total employment is a snapshot. Understanding who experiences a change also requires following those who enter, those who leave and where they go. In the Revelio study, the call-center entry rate in the twelve countries falls from 25.1 to 17.2 per hundred workers relative to November 2022; the exit rate falls from 22.3 to 18.5. The sector contracts because entries slow faster than exits, rather than because separations increase.

This distinction changes the story. A termination letter identifies a person and a date. A hire that never happens instead leaves consequences scattered across applications, waiting and interrupted pathways. This is an interpretation of the flows, rather than testimony from an interviewed applicant: the available data cannot attribute those experiences to specific individuals.
The Financial Times covered the study on October 8 in The AI Shift, describing an analysis of millions of headcount records and professional profiles across dozens of countries. This provides journalistic corroboration of Revelio’s work, rather than a second independent measurement. Professional profiles are also different from a universal administrative census: that 5.3% must be read within the dataset’s coverage.
AI is a hypothesis to test
The link with artificial intelligence remains the most delicate part. Wilkie-Rogers also considers the correction in hiring after the pandemic and cost cutting. The divergence between call centers and other office jobs makes a contribution from automation plausible, but the study does not observe which companies have replaced people with chatbots. Coinciding in time is insufficient to identify a cause.
This limitation is useful for readers. It prevents a sectoral signal from becoming a prediction that any job performed at a computer will disappear. It also prevents the opposite mistake: looking only for layoffs explicitly attributed to AI and concluding that nothing is changing when those are absent. Hiring decisions can alter an occupation before the change has a recognized name.
A stronger test would connect the adoption of tools, automated activities and headcount within the same company, comparing them with alternative conditions. Without that step, we can describe a trend and discuss its implications; we cannot assign a precise share of missing jobs to a technology. That is the difference between a documented warning and a certainty that today’s sources do not provide.

New jobs do not mean an automatic transfer
The most reassuring promise is that new jobs will absorb people released from old ones. The transition findings complicate that promise: only 10.8% of observed moves after a call-center job lead to technical support, customer success, or software and data. Almost half lead to other customer service, sales, or office and human-resources work. Growth elsewhere is not, by itself, a bridge.
Here, research by the International Labour Organization helps explain what is at stake. Its March 17 working paper, published before this week’s news, analyzes exposure to GenAI in 135 countries. In less wealthy settings, exposed office roles can represent a small part of total employment while providing an important route into formal, better work, especially for women and young people.
The report describes the risk that this route narrows before an equivalent one exists. It also warns that the same occupational title can conceal different tasks across countries, and that digital access conditions the benefits. It therefore does not justify automatically applying a global result to Italy. It offers a question: which experiences actually enable someone to move from an entry role to a more qualified one?

For someone considering retraining, the distinction is practical. Completing a course and securing a new job are two different outcomes. Assessing a pathway would require knowing who was admitted, which skills they acquired and how many subsequently found a relevant role. This is an editorial criterion for reading promises, rather than a finding from the study or a guarantee about any particular course.
Who builds the next step?
Ravinder Tulsiani, a professional in corporate learning, commented on the research on October 8, directing the question toward employers. If a company reduces recruitment into entry positions, it must check whether those positions also supplied its future specialists and supervisors. Today’s potential savings may entail a cost to find experience tomorrow. This is his interpretation, rather than an effect already measured by the dataset.
The proposal is to examine actual transitions: which roles experienced staff come from, how long it takes them to become independent, and which internal pathways work. Where a channel produces useful skills, reducing it requires an alternative. Where it does not work, preserving it out of habit does not solve the problem. Supervised training, internal mobility and hiring experienced people are choices to compare with the work that remains.
The point is not to preserve every repetitive activity. It is to understand who pays for the time needed to learn the next one. A system can answer requests more quickly and offer a better service; that benefit does not yet tell us how opportunities will change for someone who would have started by answering those requests.
The call-center story therefore also concerns people who do not work in one: it shows why technological growth and access to employment must be observed separately. In the next data, the decisive signal will be whether people reach the roles that are growing. While that transition remains rare, the visible problem is not just the job that disappears. It is the first step that goes missing.
