The jobs generative AI is supposed to hollow out are the same jobs employers are paying the most to fill. New data from Indeed Hiring Lab, published September 17 and widely circulated over the weekend, shows advertised salaries in the most AI-exposed US occupations have climbed about 46% since 2021, compared with 41% for moderately exposed roles and just 25% for the least exposed. Over the same stretch, US employers have blamed AI for 116,175 announced job cuts this year alone, roughly one in five of every layoff on the books.
Both things are true at once, and the reason they can coexist is the most important finding in the report.
Indeed economist Jack Kennedy sorted millions of salaried US postings into terciles using the company’s GenAI Skill Transformation Index, which scores how much of an occupation’s skill set could be performed or reshaped by generative AI. The high-exposure bucket holds software development, IT systems and support, data and analytics, marketing, and banking and finance. The low-exposure bucket is nursing, home health, food service, cleaning and manufacturing. Kennedy then tracked median advertised pay in each group before and after ChatGPT’s late-2022 launch.
“Advertised wages in the occupations most exposed to AI aren’t falling behind,” Kennedy wrote. “Instead, they’re growing faster than in less-exposed jobs.” The two lines tracked closely for the first year after ChatGPT, then diverged around 2024 and have kept widening since, roughly when postings in exposed occupations began to rebound from their 2022-to-2025 slump.
The headline number is partly a composition effect, and Indeed does not hide it. In the most-exposed occupations, the entry-level share of salaried postings collapsed from 29% to 10% between 2021 and 2026, while the senior share jumped from 22% to 47%. Fewer junior listings and more senior ones push a median up on their own. So Kennedy ran a difference-in-differences model to strip that out. Controlling for occupation mix, the post-ChatGPT premium for AI-exposed roles is 5.7%. Comparing each job title only against its own history, the premium is 4.7%. Holding the seniority mix constant within each occupation, it shrinks to 2.4% and loses statistical significance.
Kennedy argues that last specification may go too far. “Since AI task reshaping is plausibly one factor behind the seniority tilt,” he wrote, “it could be argued that this control may tend to over-correct, removing some of what AI is doing rather than a distortion in the data.” If AI is the reason employers now want fewer juniors and more seniors, controlling away the seniority shift controls away part of AI’s effect.
The seniority split is where the two halves of the paradox meet. By mid-2026, senior pay in exposed occupations was up 45% from 2021 versus 28% in less-exposed work, a 17-point gap. Mid-level roles show a 12-point gap. At entry level, the gap is two points. Indeed calls the split suggestive rather than conclusive, but the direction matches everything else in the data: the market is paying up for people who can supervise, integrate and direct AI systems, not for people whose first-year tasks AI can now do.
The layoff side of the ledger comes from Challenger, Gray & Christmas. Through August, AI was cited in 116,175 announced US job cuts, about 22% of the 529,914 total, making it the leading stated reason year to date. It topped the monthly list from March through July before falling to fourth in August with 3,462 cuts. Technology, the sector that overlaps most with Indeed’s high-exposure bucket, accounts for 155,126 cuts this year, 29% of all announced layoffs and up 52% from the same period in 2025.
Andy Challenger, the firm’s chief revenue officer, put the mixed picture plainly in the August release. “What we’d like to see with low layoffs is an increase in hiring activity,” he said. “While companies are making plans to hire more workers than last year, according to our numbers, it doesn’t appear those positions are being filled quickly.” Announced hiring plans through August reached 119,825, up 37% from a year earlier.
Why it matters
The dominant narrative of the past two years has been that AI exposure means wage pressure, and the Indeed data flatly contradicts it at the occupation level. But the report’s own decomposition shows the truth is narrower and, for a lot of workers, harsher: AI is not lowering the price of exposed occupations, it is changing who gets hired into them. A senior data engineer or a marketing lead who can run AI-assisted workflows is scarcer and more valuable than in 2021. A new graduate who would once have learned the trade by doing the routine work AI now handles faces a market where entry-level listings have dropped by two-thirds as a share of postings.
That reconciles the paradox. Companies citing AI when they cut are mostly shedding roles at the routine end of exposed occupations, or consolidating teams. Companies raising advertised pay are bidding for the people who can make the smaller team work. Both are the same reorganization seen from different ends. Kennedy’s conclusion, that AI has so far acted “more as a complement to skilled workers than a replacement,” is accurate and also incomplete, because a complement to experienced workers can be a substitute for inexperienced ones. The Challenger numbers deserve the same care: an employer citing AI is making a statement about strategy as much as about cause.
What to watch
The next test is whether the entry-level gap closes or hardens. Indeed’s data runs through mid-2026, and postings in exposed occupations are rebounding; if that rebound starts adding junior listings again, the pipeline problem eases. If it stays concentrated in senior and AI-titled roles, as 71% of the past year’s software-development growth has been, the pay premium and the layoff tally will keep rising together, with the market pricing AI skills at the top while quietly removing the rungs below. Watch the September Challenger report in early October, and whether AI resumes its place as the top cited reason for cuts once the August lull passes.
“Advertised wages in the occupations most exposed to AI aren’t falling behind. Instead, they’re growing faster than in less-exposed jobs.”— Jack Kennedy, Economist, Indeed Hiring Lab