Artificial intelligence is beginning to show measurable effects on parts of the labour market, with software, finance, legal services, customer contact and some creative roles among those most exposed, according to BBC News.
The report says major companies are investing heavily in AI tools that may either automate tasks or help workers perform them more efficiently. Some employers are weighing whether duties should be carried out by new staff or by AI agents designed to complete specific work.
One benchmark cited by the BBC shows rapid progress in large language models used for software tasks. Three years ago, such systems could reliably handle work that took people seconds or minutes. More recent models are increasingly able to complete more complex tasks that would take a human an hour or longer, including finding issues in a cryptocurrency contract and improving aspects of a model.
The BBC says similar trends, though at an earlier stage, are appearing in financial analysis, early legal work and some entry-level creative jobs.
Research from Stanford University, based on US wage and employment data, found a 2.7% fall in employment for 22 to 25-year-olds since ChatGPT became widely used. In the most AI-exposed sectors, including finance, software and creative industries, the fall was put at 12.8%. The BBC notes that some economists dispute whether AI is the main cause, pointing to other factors such as higher interest rates.
Online job adverts have also changed since the arrival of ChatGPT and other large language models. The BBC cites OECD analysis comparing sectors more exposed to AI, such as telemarketing and legal services, with less exposed areas including construction, cleaning and food preparation. It says the UK appeared notably affected on this measure, with its large service sector leaving it exposed to possible AI-related job losses.
The report also highlights uncertainty over the cost of AI use. AI activity is measured in tokens, small pieces of text used by systems to process and generate language. Although token costs have fallen, usage has risen sharply in 2026, with some large companies running up high bills and beginning to restrict access to advanced models.
The BBC says this may indicate limits to how much work can be automated if AI agents prove more expensive than human workers for some tasks. It also notes that some firms are turning to cheaper AI systems derived from Chinese models made freely available, adding further uncertainty to how the market may develop.
