Will AI lower wages for college graduates? The economists who track the labor market for a living now lean further toward yes. Indeed Hiring Lab’s Q3 2026 Labor Market Outlook Survey, which polls more than 120 US academics and economists every quarter, found that expectations of AI pushing down pay for college-educated workers moved more than any other reading in the survey.
Here is the awkward part. Indeed’s own job postings data, published a week earlier, shows the jobs most exposed to AI advertising pay that has climbed faster than everything else since 2021. Both findings are real. They describe different people. If you have a degree and you are looking for work this fall, the gap between those two numbers tells you a lot about where to aim.
What 120 economists said about AI and college-educated pay
The survey uses a diffusion index for AI’s expected effect on wages over the next 12 months. It runs from 0 to 100. A reading of 50 means respondents see no pressure in either direction; below 50 means they expect AI to push pay down.
For college-educated workers, that index fell from 47.3 last quarter to 42.4. Indeed’s authors, Laura Ullrich and Svenja Gudell, called it “the most pronounced quarter-over-quarter change in the entire survey.” Three months ago economists already leaned slightly negative. Now they lean clearly negative.
For workers without a college degree, the same reading sat unchanged at 50.9. Basically neutral. The economists do not expect AI to do much to their pay, up or down.
The displacement question tells the same story. A separate index on whether AI has made job loss more likely came in at 55.6 for college-educated workers, meaning more respondents think that risk has risen than fallen. For non-college workers it was 48.1.
On employment overall, the panel is mild rather than alarmed. 45% expect AI to have a minor negative effect on jobs, 25% a minor positive one, and 21% no net effect at all. So nobody in this group is forecasting a collapse. What they are forecasting is a slow squeeze on the wages and job security of people with degrees, which happens to be most of the people reading career advice on the internet.
Will AI lower wages for college graduates if AI-exposed pay is rising?
Now the other dataset. In “AI Exposure Isn’t Squeezing Advertised Pay in the US, It’s Boosting It,” Indeed economist Jack Kennedy compared advertised salaries across occupations by how exposed they are to AI. Since 2021, advertised pay in the most AI-exposed occupations rose about 46%. In the least-exposed, it rose 25%. The gap opened up noticeably around 2024.
The high-exposure group is mostly white-collar degree territory: Software Development, IT Systems & Support, Data & Analytics, Marketing, and Banking & Finance. The low-exposure group includes Nursing, Personal Care & Home Health, Food Preparation & Service, Cleaning & Sanitation, and Production & Manufacturing.
So AI exposed jobs pay more, and the premium has grown. How does that square with economists bracing for AI to pull degree-holders’ wages down?
Kennedy’s breakdown answers most of it. Once you control for which occupations are in the mix, the post-ChatGPT pay premium for AI-exposed work shrinks to 5.7%. Comparing postings with the same job title, it is 4.7%. Hold seniority constant and it drops to 2.4%. And the premium widens as you go up the ladder: large for senior roles, moderate at mid-level, negligible at entry level.
Put the two reports side by side and the picture gets sharper. The pay gains in AI-exposed fields are real, but they are piling up at the top, among experienced people whose judgment and context are harder to automate. A recent graduate applying for an entry-level analyst or junior developer role is in an AI-exposed occupation without getting much of the AI-era raise. That is exactly the group the economists are worried about.
Also worth keeping in mind: advertised pay is what employers post, and the survey is about what economists expect next. One looks backward at listings, the other looks forward at the whole workforce. Neither is a paycheck. But they point the same way for the entry level.
Where the job market outlook for 2026 points
The broader backdrop is steady and cool. Indeed’s read of the September jobs report, titled “Steadiness Without a Spark,” notes that employers added 29,000 jobs in September and unemployment ticked up from 4.1% to 4.2%. Labor force participation was 61.8%. Cory Stahle summed up employer behavior as “holding on to the workers they have and holding off on hiring the ones they don’t.”
There is a small bright spot. Job postings on Indeed fell to their 2020 baseline earlier this year, then crept back above it, and recently logged their first year-over-year gain since late 2022. The survey panel expects the Indeed Job Postings Index to inch up 0.13% in the fourth quarter and end roughly 0.89% lower by September 2027. Mean expectations for unemployment are 4.23% at the end of Q4 and 4.32% by September 2027, with the most pessimistic respondents seeing 5.5%.
That is a flat market. Hiring is not frozen, but openings are not multiplying either, so each one draws a crowd.
Jobs expected to grow in 2026, and where experts split
The survey also asked where postings will rise and fall over the next 12 months. The biggest expected gains are in Personal Care & Home Health, followed by Nursing. The biggest expected declines are in Administrative Assistance and Software Development.
Software Development shows up on both lists. Economists genuinely disagree about it. The same split shows up in the AI-specific question: the fields expected to lose the most jobs to AI are Administrative Assistance, Software Development, and Data & Analytics, while the fields expected to gain the most from AI are IT Infrastructure Operations & Support, Software Development, and Data & Analytics.
Care and health roles appear on neither AI list. They are growing for reasons that have little to do with AI, and AI is not expected to change that much.
If your degree points toward software or data, the honest read is that nobody knows how those fields net out. Some teams will hire fewer people. Others will hire more people who can work with AI tools. Which kind of team you land on matters more than the field label.
What the AI impact on salaries in 2026 means for a degree-holder
The AI pay premium rewards seniority. If you have several years in a field, especially in software, data, IT, marketing, or finance, the market is paying for your experience more than it did in 2021. Price yourself accordingly. Look at senior postings in your field before deciding what to ask for.
Entry-level degree-holders in AI-exposed fields are on the wrong side of both reports. They do not get much premium, and economists expect AI to squeeze them hardest. If that is you, aim for a role where the work involves judgment or ownership of something, so you look less like the task AI is absorbing and more like the person checking its output. “Any job in data” is too wide a target.
Growth outside AI’s reach is concentrated in care and health. That is useful for some career changers and irrelevant to others. Do not pivot into nursing because of a survey. Do notice that the jobs economists feel most confident about are the ones furthest from a keyboard.
A few moves follow from this. When you search, filter by seniority as well as title. A posting that reads like a junior role with a senior title, or the reverse, gives itself away in the pay band, and the premium lives in roles with real scope.
Put AI-adjacent work on the page. Kennedy’s data shows a premium of 4.7% even between postings with the same job title, so something beyond the title is separating the better-paid listings. Concrete examples of using AI tools to ship work or check it are the closest thing to that signal you can show.
And treat software and data as team-by-team markets. Economists cannot agree on the field as a whole, so read the individual company instead. A team that is shipping and growing is a different bet from one that is cutting, whatever the sector average says.
Why direct outreach matters more in a steady-but-cool market
A market adding 29,000 jobs a month, with employers holding on to the people they already have, is a market where open roles stay scarce and applications stack up. AI makes that worse in a specific way: writing and sending an application now costs almost nothing, so every posting gets more of them. Portals sort that pile by keyword match. The things that justify a higher salary in an AI-exposed field, like judgment, scope, and evidence that you can use the tools well, are hard to convey to a filter.
They are easy to convey to a person. The hiring manager for a senior data or engineering role is the one deciding whether your experience is worth the premium Kennedy measured. A short, specific message to that person (“saw the team is rebuilding its forecasting pipeline; here is a similar project and what it cut”) does more for your offer than a perfectly tuned resume sitting deep in an applicant tracking system. The math on volume versus aim is covered in how many jobs to apply to, and the conclusion is the same here: fewer applications, better aimed, with a human on the other end.
For entry-level degree-holders, outreach matters even more. If the posting itself is the kind of work AI is absorbing, the only way to show you are more than that is to talk to someone who can see it.
The takeaway
So, will AI lower wages for college graduates? The economists surveyed by Indeed Hiring Lab increasingly think so, and the postings data says the pressure lands hardest at the entry level, while senior people in AI-exposed fields are still getting paid more. The job market outlook for 2026 is flat, and flat still means people are getting hired. In a flat market the winners are the people who aim at roles where their experience earns the premium, and who get in front of the person making the call instead of waiting in the queue.
Reaching out to hiring managers works. The hard part is the research: figuring out who they are, finding something worth saying, and writing a message that does not sound generic. angld.AI handles that pipeline. Paste a job posting, and it identifies the decision maker, researches them, and drafts a personalized outreach message in about 60 seconds.
