The Class of 2026 already knows AI, but employers are still trying to teach them

Class of 2026 graduates have spent their entire college careers with generative AI as a default tool. They mention it on their resumes nine times more often than the Class of 2022 did, and most of those mentions come from students who didn't major in computer science. They are also the most pessimistic cohort about their job prospects in years, walking into a difficult entry-level market and an interview process that has broken on both sides.

This combination is rarely framed as a paradox in the coverage. The Class of 2026 is usually positioned as the cohort most at risk from AI – disrupted, displaced, replaced before they begin. But four labor economists who gathered at a recent Deel-hosted event in New York City laid out a different read of the data. This cohort knows AI well, but the employers interviewing them don't yet recognize how well.

The most AI-fluent cohort ever to enter the workforce

Randy Tarnowski, Head of Economic Research at Handshake, presented data on the Class of 2026 that goes well past the standard early-career AI mention. Their resumes mention AI skills more than nine times as often as Class of 2022 resumes did at the same career stage, with two-thirds of those mentions coming from students outside computer science majors, according to a Handshake report on the cohort.

The timing helps explain the depth of the fluency. The Class of 2026 was in the junior year of high school when Covid forced learning online, then started college around the time ChatGPT launched in November 2022. They have spent their entire undergraduate experience with generative AI as a default tool, used across coursework regardless of major. "They've gone through a lot," Tarnowski said.

Employer demand has begun to catch up to the supply, but not at the same pace. Deel's chief economist Lauren Thomas pointed to data showing that the share of active job descriptions mentioning AI-related skills has more than doubled since January 2023, from 2.3% to 5.6%. For roles targeting 18-to-24-year-olds, the share has nearly tripled. The gap between what this cohort brings and what employers are asking for is closing, but slowly.

Being interviewed by people who don't yet see it

The employers hiring this cohort still default to AI as a senior-level differentiator, a skill set built up later in a career, rather than a baseline competency that today's graduates arrive with. Tarnowski described the mismatch directly: "Freshmen and CEOs have access to the same tools."

What employers don't yet see, in his framing, is that this group is already operating with the tools rather than learning them on the job. "Entry-level workers are at the tip of the spear with these tools," he said. "Early talent is often overlooked as skilled AI users."

Tarnowski's research also found that AI usage is the strongest single predictor of optimism about the job market – stronger than any demographic variable. Combined with the data on resume mentions, the cohort that has spent the most time with these tools is also the one most ready to use them in a role, but they are competing for jobs in a market that still treats AI fluency as a bonus rather than a baseline.

Meanwhile, the cohort itself is feeling worse about its prospects. Glassdoor's Employee Confidence Index hit a record low in April 2026, with only 43.8% of employees reporting a positive six-month business outlook – the weakest reading since the index began in 2016. Entry-level workers tracked below the overall index. Daniel Zhao, Chief Economist at Glassdoor, said the sentiment is real but not exclusively about AI: "Younger workers are nervous about where the market is at. [This is] partially driven by AI, but not totally." Glassdoor's analysis attributes much of the drop to the current geopolitical climate and the resulting spike in energy prices.

What graduating into a frozen market actually means

Perception isn't the only challenge this cohort faces, as graduating into a difficult market has measurable consequences. Zhao framed it as primarily a matter of timing and luck. The job market a person graduates into has a huge effect on their career – and to some extent, he noted, that's based on luck. He described new grads in the past one to two years as being in a rough situation through no fault of their own.

Tarnowski pointed to the 2008 cohort as an example. Workers who graduated into the financial crisis carry quantifiable lifetime disadvantages – they did not recover at the rate economists would have expected. The research on past recessions is consistent: people who graduate into bad markets struggle for longer than the market itself stays bad.

Tarnowski added a complicated framing to that history. Past research, he noted, suggests that during transformational technology shifts, younger workers actually adapt faster than older workers because they aren't holding on to legacy skills. "I actually think that younger workers have a lot to bring that, if you're an older worker, you actually might struggle with," he said.

The current data on this cohort reflects both pressures at once. Thomas's research shows under-25 job hopping has dropped by more than half – the steepest decline of any age group. With fewer entry-level openings being created by mid-career turnover above them, this cohort is more stuck inside its first roles than its predecessors were.

What companies should actually be hiring for

The conversation at the dinner kept circling back to what employers should actually be evaluating in early-career candidates if AI fluency is now table stakes. Guy Berger, Workforce Economist in Residence at Guild, made the case that the bigger educational problem isn't whether new graduates can use AI – it's whether the system is set up to keep them current as the tools evolve.

He gave the example of software developers, whose pre-2022 skill set lost value dramatically once coding agents could write production-grade code on their own. "Writing out code on your own is now not a skill that's very valuable, whereas being able to curate code that's written by a coding agent is really valuable," Berger said. His framing for the educational system is that degrees should function as a platform on which workers can "bolt on" learning after the fact, rather than "one big chunk that is going to be your meal ticket for the rest of your life."

He pushed the framing further to the next cohort behind this one. "Using AI for 22-year-olds in 2030 will be like, 'Do you know how to use a cell phone?'" he said. The implication for the Class of 2026 is that the advantage they have is real but has a shelf life. Within a few years, AI fluency will be baseline for every cohort entering the workforce. Until then, this cohort is uniquely positioned – and after that, the differentiation that lasts will sit elsewhere.

Zhao pointed to a different dimension that doesn't get measured well: emotional intelligence, particularly the kind picked up by being physically present at work. Glassdoor's research found that expectations of managers have shifted significantly since the pandemic, with employees increasingly evaluating managers on emotional intelligence, expectation-setting, and protection against burnout. Zhao said this extends down to entry-level workers as well, where there is more of a baseline expectation around how people should behave in the workplace.

He flagged a complication for the current cohort, in that remote work makes it harder for entry-level workers to pick up the institutional knowledge that mid-career workers absorbed by sitting in offices. Things as small as "is it weird to interact at the coffee machine in this way?" are observational learnings that don't transfer over Zoom. Interns and entry-level workers actually show more of a preference for in-person work than the broader workforce does, he noted. It's mid-career workers who, having already absorbed that workplace fluency, are the most resistant to returning to the office.

Thomas added a longer arc framing, sharing that educational qualifications have steadily expanded for over a century, with each generation asked to bring more credentials to the labor market than the one before. She predicted the trend continues, with more emphasis on co-ops and internship experience. "If I were running a university, I would be trying to invest in these programs as much as possible," she said.

The long view

The cohort entering the workforce this year is walking into the most difficult entry-level market in years, with measurable disadvantages from the timing alone that won't fade quickly. But the AI fluency they have built – often messily, often outside the curriculum, often in majors that don't show up in any technical category – is a different kind of asset than what most prior cohorts brought with them. It is the kind of asset that compounds with use.

Employers treating the Class of 2026 as a group that needs to be taught AI are misreading the data. The cohort that will look most valuable in five years – once AI fluency becomes baseline rather than differentiator – will be the one that knew the tools first. They are already in the room.