AI is changing the workforce, but its impact on jobs is different for each industry.

Will AI replace your job? The answer will depend on your industry, what you do and how quickly businesses pick up on new technologies. Many technology companies are saying artificial intelligence will significantly improve productivity, but economists and labour market experts say the long-term impact is uncertain.

The companies that build tools for artificial intelligence keep talking up their ability to automate more and more job functions. Many companies will probably opt for a mix of automation and human supervision instead of replacing workers entirely, enabling employees to concentrate on more meaningful work as AI takes care of repetitive or routine tasks.

As companies pursue efficiency and cut operating costs, investment in AI has gained momentum. Executives are now more often talking about the possibility of keeping a steady workforce while at the same time ramping up the use of AI-driven software, including so-called “AI agents” that can perform specific tasks with little human oversight.

Economists warn the shift could have major implications for workers if governments and companies don’t prepare. Nobel Prize-winning economists have called on policymakers to ensure that AI boosts productivity and living standards without causing widespread job losses. Meanwhile, in the UK, employers have reported increasing skills shortages as the advent of AI shifts hiring needs.

AI capabilities will only improve

In the past three years, large language models (LLMs) have shown rapid progress in capabilities.

Industry benchmarks indicate that AI systems have progressed from doing work that previously took seconds or minutes of human time to handling larger jobs that skilled workers might spend an hour or more on. For example, in software development, some models are able to identify vulnerabilities in the code, help debug and make functional software more efficiently than previous generations.

Similar trends are emerging in financial analysis, routine legal research, customer support and some entry-level creative work. Experts warn, however, that AI performance is highly variable depending on the complexity of the task.

Which occupations are most exposed?

Studies have found that jobs that require digital, administrative or knowledge work are more likely to be disrupted than those that require physical labour or customer interaction.

A Stanford University review of employment and wage data found that employment among workers 22 to 25 years old fell 2.7% after the widespread use of ChatGPT. Jobs thought to be highly exposed to AI – such as finance, software development and creative industries – saw a fall of 12.8%.

But economists are wary of treating the figures as proof that AI was the only cause of the shifts. Higher interest rates, slower hiring and the broader economic picture may have also played a role.

Jobs in healthcare, childcare, personal care services and other hands-on fields are less likely to be affected by automation because they require a great deal of physical presence, interpersonal skills and complex human judgement.

New directions in recruitment

Furthermore, data from job postings indicate that AI could be affecting the way companies are hiring.

The Organisation for Economic Co-operation and Development (OECD) analysis shows that vacancies for occupations with higher AI exposure, such as legal services and telemarketing, have fallen more sharply than in sectors with lower exposure, like construction, cleaning and food preparation.

The United Kingdom is especially exposed because of the size of its service-sector workforce, though analysts say hiring activity is also affected by broader economic conditions.

AI use on the rise = costs on the rise

Fast forward to 2026, and the use of AI in business has exploded. Organisations are processing huge amounts of AI “tokens” – the units of text that AI models use to process information and generate responses.

Many companies initially encouraged employees to maximise their use of AI in search of productivity gains. But the widespread use of sophisticated AI tools has also generated significant operational costs, prompting some companies to restrict staff usage.

The experience shows that, although AI can automate some tasks, it may not always be economically practical to fully replace human workers. “Saving money by reducing staff may not be enough to offset the costs of running the advanced AI systems,” he said.

Meanwhile, companies are experimenting more and more with lower-cost open-source and commercial AI models, including those from Chinese developers, which could potentially reduce deployment costs and increase adoption.

A questionable transition

Evidence continues to emerge, but some trends are becoming clearer. AI is taking over more of the work that used to be done by knowledge workers, particularly in software development, finance, legal services and customer support.

But researchers emphasise that the technology is more likely to transform many jobs than to eliminate them. The labour market will evolve over the next few years depending on the speed of adoption, the regulatory response, economic conditions and the cost of AI systems.