Early data show AI is changing some jobs but not total employment

In short: Research so far suggests AI is shifting work in specific jobs and age groups, but it has not clearly changed total employment across the whole economy.

What's going on

Several studies looking at the US economy from 2010 to 2023 find no big, clear change in total employment that can be pinned on AI. Some jobs that use AI heavily have shrunk, but other jobs have grown, and some companies that adopt AI have expanded. Yale’s Budget Lab also says the overall mix of jobs still looks stable, and early measures of “AI exposure” do not line up neatly with changes in employment or unemployment.

At the same time, some groups appear to be getting hit. A Stanford working paper cited by the Carnegie Endowment found that early-career workers ages 22 to 25 in the most AI-exposed jobs saw a 16% relative drop in employment after ChatGPT arrived. MIT Sloan research suggests the impact depends on how much of a job AI can do. If AI can handle most tasks in a role, jobs for that role inside a company fall by about 14%, but if AI only helps with a few tasks, hiring can rise as workers focus on what AI is bad at, like judgment and creativity.

Other data points lean more positive. An analysis of about 1 billion job postings, from PwC and summarized by LinkedIn, found wages and job numbers rising in some roles that people assume are easiest to automate, including customer service and software work.

What to watch

A big reason the story looks messy is timing and measurement. Layoffs show up fast, but job creation can take longer, like a small business slowly hiring as it grows. Watch for better tracking of which tasks are changing inside jobs, and whether younger workers in exposed fields continue to face a tougher entry-level market.

Source: NYTimes