← HR Magazine

Practical HR for companies

Who Loses Their Chair? The Jobs AI Will Eliminate, Transform and Create

AI is unlikely to end work, but it will reduce routine office tasks, redesign professional roles and make entry-level hiring harder. Which jobs face the greatest risk, and how can people prepare?

Trautenberk Consulting Editorial Team 3. 9. 2026

AI will not take everyone’s job. But it will take over a significant part of some jobs

The idea that artificial intelligence will “end work” within a few years is exaggerated. Yet it is equally unconvincing to claim that nothing fundamental is happening. Generative AI can already produce text, summarize documents, translate messages, prepare tables, write simpler code, propose visual concepts and handle parts of customer communication in seconds. It does not automatically threaten whole occupations. It targets tasks – and the way those tasks are bundled determines how many people a company will need.

International organizations use different methods, so their estimates vary. In 2025, the ILO and Poland’s NASK institute estimated that generative AI could affect around one quarter of jobs worldwide and 34% of jobs in high-income countries. They stress that job transformation is more likely than complete replacement. The International Monetary Fund, using a broader definition of exposure, puts the figures at roughly 40% of jobs globally and 60% in advanced economies.

Exposure is not the same as elimination. It means that an important share of tasks can be automated or accelerated. The outcome will depend on the cost and quality of the technology, regulation, customer trust and whether organizations can redesign their processes.

Which jobs are likely to decline most

1. Administrative and clerical routine

The greatest pressure will fall on jobs built mainly around moving information from one format to another: data entry, transcription, basic record keeping, document sorting, standardized forms, simple reports, calendar management and routine email communication.

The World Economic Forum lists postal clerks, bank tellers, data-entry clerks, cashiers, administrative assistants, secretarial roles, and accounting and payroll clerks among the fastest-declining roles through 2030. Not all of these occupations will disappear, but companies may handle the same workload with fewer people.

2. Basic customer support and telemarketing

First-line support based on repetitive questions is a natural environment for AI: requests can be classified, answers retrieved from a knowledge base and conversations summarized automatically. Scripted sales calls and basic outbound contact are similarly vulnerable.

This does not mean the end of human customer service. People will handle escalations, conflict, complex cases and situations requiring empathy or commercial judgement. Purely scripted roles, however, are likely to shrink.

3. Basic production of text, graphics and translations

AI reduces the cost of a first draft. The most exposed work is therefore interchangeable production: product descriptions, routine social posts, simple banners, image variants, literal translations and content produced primarily for volume. WEF now also includes graphic designers among the faster-declining roles.

The need for communication, design and localization will not disappear. What will decline is the willingness to pay a person merely to generate version one. Value will shift toward concept, brand, editorial judgement, fact-checking, audience understanding and accountability for the result.

4. Parts of accounting, legal and analytical support

Document comparison, data extraction, checks of standard contract clauses, research, recurring reports and initial transaction classification are structured tasks. Qualified professionals will remain, but teams may need fewer people for preparatory work.

The risk is especially important for junior roles traditionally built around searching, copying and preparing background materials. This creates a difficult question for employers: where will future senior professionals gain experience if software performs much of the work that once trained them?

5. The most automatable parts of programming

AI can generate common blocks of code, tests and documentation and correct simpler errors. The OECD Skills Outlook 2025 points to disruption for application programmers from low-code tools and generative AI, while also expecting a stronger premium for system architecture, cybersecurity and complex problem solving.

Programming will not disappear. The role under pressure is the person who only converts a precise specification into standard code without understanding the wider system, the user or the business objective.

Which professions will be transformed rather than eliminated

HR and recruitment

AI can draft job advertisements, identify candidates, summarize CVs, suggest interview questions and process interview notes. Recruiters will move away from administration toward diagnosing business needs, assessing motivation and potential, advising hiring managers and building trust. At the same time, responsibility for bias, data protection and explainability will increase.

Marketing, media and communication

Producing variants will become faster and cheaper. Yet the amount of average content will grow, increasing the value of original insight, credibility, distribution and customer knowledge. Successful specialists will not compete with AI on typing speed. They will manage the complete system: brief, data, tools, editorial judgement, verification and measurement.

Law, finance and consulting

AI will accelerate research, comparisons, modelling and document preparation. Professionals will spend less time producing background materials and more time interpreting uncertainty, explaining consequences and accepting responsibility. The advantage will belong neither to advisers who ignore AI nor to those who trust it blindly, but to those who can verify its output and apply it to the specific situation.

Software development and technical professions

Developers will increasingly act as architects, integrators and reviewers of machine-generated code. Syntax will still matter, but it will not be enough. System design, security, testing, data skills and an understanding of the real-world problem will become more valuable.

Healthcare, education and care work

AI can support documentation, preparation, personalization and decision-making. The core of the job, however, involves relationships, physical intervention, trust and responsibility. A doctor, teacher or social worker may become more productive with AI without becoming automatically replaceable. In these occupations, a redesigned working day is more likely than job extinction.

What early real-world evidence shows

One of the most widely cited field studies followed 5,179 customer-service agents. The NBER study found that a generative assistant raised productivity by 14% on average and by 34% among less experienced and lower-performing workers. AI did not immediately replace people. It transferred some of the practices once concentrated among the best colleagues to a wider group.

At the same time, there is a warning signal at the beginning of careers. A revised 2026 study from Stanford Digital Economy Lab does not find an economy-wide employment collapse, but employment among 22–25-year-olds in the most AI-exposed occupations was 19% below the pace implied by less-exposed peers. The result is descriptive rather than definitive proof of causation. It nevertheless suggests that organizations may first reduce junior hiring rather than dismiss all experienced workers.

European vacancy data also point to a change in the composition of demand. In 2026, Cedefop reported post-generative-AI declines in the share of vacancies for software development, sales and marketing and some information roles, while shares rose for technicians, mechanics, construction and transport occupations. Several factors affect these correlations, but the direction fits the technology: digital routine is easier to automate than work in an unpredictable physical environment.

Which new jobs will emerge

Most new opportunities will probably not have “AI” in their job title. The technology will create several visible specialties, but its greater effect will be to add new responsibilities to existing professions.

AI product managers and process architects will select viable use cases and connect technology with real operations.
AI governance, audit and compliance specialists will manage risk, documentation, copyright, data protection and regulatory requirements.
AI security specialists and red teams will test model manipulation, data leakage and misuse of automated systems.
Knowledge curators and data stewards will ensure that corporate assistants use current, authorized and high-quality sources.
AI evaluators and output reviewers will build tests, monitor error rates and determine when a person must take over.
Domain specialists augmented by AI – recruiters, lawyers, engineers, salespeople and physicians, for example – will gain an advantage by combining deep expertise with the ability to direct new tools.

WEF expects the combined impact of technological, demographic, economic and other trends to create 170 million jobs and displace 92 million by 2030, a net gain of 78 million. This is not a forecast of AI alone, and new jobs will not automatically appear in the same place or for the same people as those lost. It does show, however, that fewer routine office jobs do not necessarily mean less work overall. Fast-growing roles include big-data, AI and machine-learning specialists, fintech engineers, software developers and security specialists, but also care, education, construction and other occupations in the physical economy.

The Czech labour market: change has already started

According to the Czech Statistical Office, 17.6% of Czech enterprises with ten or more employees used at least one AI technology in 2025, compared with 11.3% in 2024 and only 5.9% in 2023. Among large enterprises, the share reached 54.1%. Thirteen percent of enterprises used AI to generate text or computer code.

This matters even to people who do not see themselves as technology workers. AI will not spread only through the recruitment of specialists. It will enter office suites, CRM platforms, accounting, recruitment systems, manufacturing applications and customer care. Many employees will encounter it as a new feature inside a tool they already use.

Should people be afraid?

Fear of an abstract robot that will take every job is not useful. Taking the change seriously is. The most vulnerable person is not someone in one particular industry, but someone whose value is based mainly on quickly performing standardized digital tasks. More resilient work tends to combine several characteristics:

it requires trust, negotiation or care for another person;
it takes place in a changing physical environment;
it carries legal, professional or managerial responsibility;
it deals with an ambiguous problem rather than only a precise instruction;
it combines domain knowledge, organizational context and sound judgement.

The impact will not be distributed evenly. The ILO reported in 2026 that, because women are strongly represented in administrative and clerical work, 29% of employment in female-dominated occupations is exposed to generative AI, compared with 16% in male-dominated occupations. Employers therefore need to plan reskilling before reorganization disproportionately affects particular groups.

How workers can prepare

Break your job into tasks. Identify activities that are repetitive, digital and easy to check. These are likely to be automated first.
Learn to use AI inside a real workflow. Prompting is not enough. The valuable skills are defining the task, using sources, checking errors, protecting data and integrating the output into a process.
Deepen your domain expertise. A person who cannot distinguish a sound result from persuasive nonsense cannot safely supervise AI.
Move closer to the problem and the customer. Understand why the work is performed, who makes the decision and what economic or human impact the outcome creates.
Build skills that complement AI. These include critical thinking, communication, negotiation, leadership, system design, data literacy and accountable decision-making.
Document outcomes. A portfolio of improved processes, saved time or better decisions will be more valuable than a general claim that you “know AI”.

What employers should do

Do not measure only the number of jobs removed. Map tasks, capacity, quality, risk and future competence needs.
Let employees co-design the change. People doing the work know where errors, delays and unnecessary administration occur.
Protect entry-level career paths. If AI takes over junior tasks, the organization must create new forms of learning, mentoring and gradual responsibility.
Set rules for data and human review. Employees need to know which tools are permitted, what data must not be entered and who is accountable for the outcome.
Invest in reskilling before restructuring. WEF reports that 41% of employers expect workforce reductions where AI can replicate work, while 77% plan to develop employees’ AI skills. The decisive question is whether training arrives before organizational change.

Conclusion: work will remain, comfortable routine will not

AI is unlikely to erase whole categories of people from the labour market. It will, however, shrink teams built around administrative production, make the first draft cheaper and increase the performance of the strongest workers. Some roles will disappear, many will merge and almost every office profession will change.

The most useful question is therefore not, “Will AI take my job?” It is, “Which part of my job will it take over, who will then carry responsibility, and what higher value will I create?” Neither workers nor employers need to panic. But they should not confuse calm with inaction.