AI at Work: Will it Augment or Replace Jobs? Examining the Most Affected Professions

Artificial intelligence (AI) companies are making significant promises about their tools' ability to replace human labor, with some jobs expected to be automated and others augmented. Leaders of major global corporations are investing heavily in AI, partly due to potential savings on personnel costs. This trend suggests a shift towards stable team sizes, as investors weigh whether tasks should be handled by new hires or by "AI agents"—virtual workers performing specific, sometimes highly skilled, functions. Economists, including Nobel laureates, have warned that immediate action is needed to ensure AI promotes living standards rather than causing widespread job displacement. Recent reports from London businesses highlight difficulties in finding necessary skills as AI transforms the labor market.

Recent data indicates that newer AI models can perform increasingly complex tasks. Three years ago, large language models (LLMs) reliably handled tasks humans completed in seconds or minutes; now, they can manage tasks taking about an hour, such as identifying flaws in cryptocurrency contracts or optimizing software models. The next generation of models may be capable of full self-development in programming within a year or so. Similar patterns, though in early stages, are emerging in financial analysis, preliminary legal work, and some entry-level creative roles. Comprehensive analyses from the U.S. show a decline in employment for young adults aged 22-25, particularly in sectors highly exposed to AI like finance, software, and creative industries, with some economists attributing this to factors beyond AI, such as rising interest rates. Online job postings have also been affected since the advent of LLMs. An OECD analysis on job postings in the UK, a country with a service-sector concentration highly exposed to AI, showed significant impacts even when interest rates were stable or decreasing, predating recent tax increases. The substantial increase in AI usage, measured in tokens, has led to massive costs for companies, prompting some to ration usage and potentially indicating limits to automation, as virtual workers might become more expensive than human ones depending on the task. Some companies are also shifting to cheaper AI forms derived from Chinese models.