AI job impact emerges in software and services

AI tools are beginning to shift from assisting with brief tasks to handling more complex work, particularly in software. Benchmarks for large language models show systems that once reliably completed only tasks humans took seconds or minutes to do can now handle work lasting an hour or so, including finding issues in cryptocurrency contracts and improving software models. Similar patterns are emerging earlier in financial analysis, legal work and some entry-level creative roles.

Labour-market evidence is still contested. Stanford University’s analysis found a hit to employment for 22 to 25-year-olds of 2.7% since ChatGPT became widespread, rising to 12.8% in the most AI-exposed sectors such as finance, software and creative industries. Not all economists agree, with interest rate rises among possible explanations. OECD jobs-posting data also shows pressure in highly exposed sectors such as telemarketing and legal services compared with less exposed work including construction, cleaning and food preparation, with the UK notably affected.

Corporate AI adoption has accelerated sharply in 2026 as companies track token use and push employees to gain productivity from advanced models. But heavy use of agentic tools has produced large bills, forcing some companies to ration access. Cheaper models derived from Chinese systems may lower costs, leaving the scale and speed of job automation uncertain.