Artificial intelligence could displace workers on a large scale, gradually reshape most professions, or create entirely new areas of employment. Which scenario prevails will depend on the pace of technological progress, the reliability of AI, and the economy’s ability to adapt. This is discussed by researcher and futurist Teddy Tavil in his article “AI’s Impact on the Labor Market: Three Views on the Future of Employment.”

In his view, AI will become one of the main drivers of labor market transformation in the coming years, although a definitive forecast is still impossible. The question is not whether AI will affect the labor market, but how quickly change will happen, how extensive displacement will be, and whether the economy can create new opportunities faster than old ones disappear.

AI could displace a significant share of office workers

According to the study, supporters of the more alarming scenario believe that over the next decade AI will master most tasks in knowledge-based professions. If companies adopt such systems quickly, demand for human workers in certain sectors could drop sharply. This view rests on three assumptions: AI capabilities will continue to grow, employers will actively replace staff, and new jobs will not offset the losses.

Infrastructure, investment in data centers, and models’ access to tools are expanding the range of tasks AI can perform. The OpenAI GDPVal study showed that across 220 professional tasks, experts preferred GPT-5.4’s answers (March 2026) to human ones or considered them equivalent in 83% of cases. Success in individual tasks is not the same as fully replacing a specialist, but quality is improving, and so are the incentives for automation.

AI works around the clock and is often cheaper than people. Entry-level positions are especially vulnerable: junior lawyers and consultants, financial analysts, office staff, executive assistants, and customer service specialists. A preliminary Stanford study (November 2025) found that employment among 22- to 25-year-olds in occupations most exposed to AI fell by 16% relative to other groups. No such effect was observed among older workers or in “safer” professions.

The main risk is that automation will outpace the creation of new jobs. Productivity will rise, but employment will not. The benefits will be concentrated among technology owners, while some people will lose income and prospects.

Why progress may not necessarily lead to rapid unemployment

The opposing view is that AI’s impact will be slower than the more alarming forecasts suggest. Even revolutionary technologies have never transformed all professions at once. There are three obstacles: model limitations, reliability, and the complexity of implementation.

Modern systems still struggle with complex multi-step tasks, sustained planning, and adapting to new situations. The Remote Labor Index (March 2026) showed that the best system reached or exceeded human performance on only 4.17% of complex freelance tasks that take humans several days to complete.

Reliability is the second barrier. AI makes mistakes with confidence. In law, finance, or medicine, the cost of an error is high, and checking results can sometimes take nearly as much time as doing the work itself. The savings end up being smaller than expected.

The third barrier is organizational. Processes must be redesigned, security ensured, responsibility defined, and employees trained. In regulated industries, strict qualification requirements apply. History shows that electricity and IT took decades to be adopted. AI, too, will need time to overcome technical and organizational constraints.

The third scenario: AI will create new professions and increase demand for people

Optimists believe that automation not only destroys jobs but also creates new ones. Rising productivity and lower prices expand markets and generate additional demand for human labor. Three mechanisms are at work.

The first is a change in the content of professions. AI takes over routine work, while people focus on judgment, communication, creativity, and decision-making. Sales systems search for clients and prepare proposals, while managers spend more time negotiating and closing deals. If companies use AI to improve quality and enter new markets, employment may hold steady or grow. A Finnish study covering 1994–2018 showed that companies introducing industrial automation increased employment by 23% through expanded production.

The second is income redistribution. Cheaper goods and services leave people with more money to spend elsewhere. Demand supports sectors where presence, interaction, and an individual approach matter: healthcare, education, construction, hospitality, and personal services. This effect is not guaranteed; if income flows only into new AI investment, the compensation may be insufficient.

The third is new companies and professions. Specialists will be needed in integration, quality assurance, risk assessment, and process redesign. AI is lowering the cost of software development and analytics, allowing smaller companies to offer services that were previously available only to large firms. New models will emerge in education, financial consulting, medical analytics, and legal services. The new jobs will not resemble the old ones, which is why they are hard to list in advance. The transition will take time and training, and some groups will still face difficulties.

The future depends on the pace of technology and the economy’s adaptation

In Tavil’s view, the three scenarios do not exclude one another. They may unfold in parallel across different industries or replace one another over time. At first, implementation may be slow because of limitations; then automation may accelerate; and in some sectors, production and new professions may grow.

The key question is not only “how intelligent AI will become,” but also how reliably it will perform real-world tasks, how quickly companies will restructure their work, and whether new forms of activity can be created.

Tavil emphasizes that governments need to monitor not only the technology itself, but also its real impact on employment — data on adoption, hiring, wages, and skills. Special attention should be paid to supporting vulnerable workers through training programs, short-term retraining, and wage insurance. It is impossible to centrally plan the entire future labor market, so the task should be framed differently: make it easier for people to transition and prepare tools in case the pace of change accelerates.

Neither mass displacement, nor smooth adaptation, nor an explosion of new jobs is guaranteed at this point. The future will be determined by a combination of technology, incentives, constraints, and the decisions of companies and governments. The debate over employment is ultimately a debate over whether the economy can create opportunities faster than technology closes them off, Tavil concludes.