In the last three years, artificial intelligence (AI) tools like ChatGPT and Claude have proliferated rapidly. Today, these technologies are able to write code, answer complex questions, and perform a variety of tasks that were earlier done by humans. A new Stanford report has thrown light on how these AI tools are impacting jobs in the United States with the help of detailed payroll data from millions of workers between 2021 and mid-2025.
The findings of the report seem to suggest that while AI is reshaping the workforce, it is doing so in an uneven manner. The report states that young or entry-level workers in certain industries are among the hardest hit, akin to ‘canaries in the coal mine’. Here’s a look at some of the key takeaways from the report.
Entry-level workers at risk
According to the report, jobs like software engineering and customer service are highly exposed to AI since most of their daily tasks, such as writing code or answering queries, can now be done by AI systems. The report shows that there has been around a 13 per cent overall decline in these jobs for workers aged between 22 and 25, with nearly 20 per cent in some occupations. On the other hand, older workers in the same fields have witnessed steady employment. In less AI-exposed jobs such as nursing, younger workers are still able to find employment. This essentially means that for fresh graduates, entering jobs involving programming or call centres has become much harder.
Jobs are growing, but not for young workers
According to the report, the US job market continues to be strong with low unemployment. However, for those aged 22-25, employment in AI-exposed jobs is flat. The report shows that between late 2022 and mid-2025, young workers in the most exposed jobs witnessed a 6 per cent decline in employment. On the contrary, older workers in the same jobs saw employment rise by up to 9 per cent. This simply means that even though the economy may look healthy, young job seekers in AI-heavy industries seem to be struggling.
The report asserts that not all AI is bad news. The researchers have drawn distinctions between AI that fully automates tasks and the ones that help humans do their jobs better. Jobs where AI automates tasks like writing code directly saw a steep decline in employment among young workers. On the other hand, in fields where AI is used as an assistant, the employment is stable or even rising.
What is the trend about?
While one may be compelled to believe that the job declines are caused by other factors such as interest rate hikes or tech industry slowdowns, researchers tested this by controlling company-specific shocks. Even after adjusting these, the employment for young workers in AI-exposed fields remained significant. Essentially, job cuts do not mean the tech industry is in decline or going through a rough patch; rather, they point to AI reshaping demand for workers.
What about salaries?
Even though there have been significant movements in jobs, wages haven’t moved much. The report shows that salaries for young workers in AI-exposed jobs are not dropping sharply; rather, companies are making adjustments through fewer hires. In simple words, companies are not slashing pay but cutting back on how many young workers they bring in. The report essentially says that AI’s first impact seems to be reduced entry-level opportunities and not falling wages for those who are already employed.
Story continues below this ad
Why are young workers hit the hardest?
According to the researchers, one reason is that AI replaces ‘book knowledge’ more easily than ‘tacit knowledge’. This means fresh graduates come with a lot of textbook knowledge and skills that AI can easily mimic. On the other hand, experienced or senior workers possess accumulated tricks of the trade, better judgement, and people skills, something which AI struggles with. This is perhaps the key reason why older workers seem to have been shielded from AI job losses so far.
The report ‘Canaries in the Coal Mine? Six Facts About The Recent Employment Effects of Artificial Intelligence’ is a research paper by Stanford Digital Economy Lab, which is part of the Stanford Institute for Human-Centred AI (HAI). The report is authored by Stanford researchers Erik Brynjolfsson, Bharat Chandar, and Ruyu Chen.
0
0
Average Rating