That’s the somewhat obvious—and yet still somewhat remarkable—takeaway from the newly released Wharton-Accenture Skills Index (WAsX), an empirical benchmark that tracks more than 150 million unique U.S. profiles and 100 million job postings. The research team, comprising Wharton professor Eric Bradlow and Accenture’s James Crowley, Ken Munie, and Selen Karaca-Griffin, found the labor market is operating in a state of structural imbalance. The signaling gap in a nutshell: Workers are overwhelmingly promoting generalist traits such as communication, leadership, and problem-solving, while employers are increasingly desperate for specialized, execution-oriented skills that remain in short supply.

“I think just that there is this massive skills gap,” Bradlow told Fortune in an interview. “The skills that people, that companies are saying they want in job postings are just not the way people are representing themselves.” The professor, who serves as the vice dean of AI and analytics at the vaunted business school, said he sees employers drowning in “table stakes” skills like leadership, teamwork, and communication—capabilities that everybody’s putting down on their résumés, which means they no longer differentiate candidates in the market. Bradlow—a statistician and computer scientist by training who has been at Wharton for 30 years; has led a data science program for 20 years; and has drilled down on AI for the past decade—added that he was certain about one thing: This situation can’t last.

The signaling trap

The research highlights an oversupply of “generalists,” with workers competing on “safe” signals that are socially reinforced, but have become so common that they no longer differentiate talent. In the WAsX “surplus-deficit map,” these broad traits appear as the most oversupplied. Conversely, the market rewards technical depth, scientific methods, and analytical precision—capabilities that workers consistently under-signal.

This means your job title and your résumé won’t protect you anymore. Once a reliable proxy for what workers do, job titles “no longer describe how work actually gets done,” the report says.

According to Bradlow, employers increasingly say, “Don’t tell me your title. Tell me what skills you have, and tell me what you actually do.” Yet the CV, which remains the “dominant signaling tool,” still tends to emphasize broad traits such as communication, leadership, and problem-solving “without really communicating what value a particular worker adds.”​

Artificial intelligence is acting as a massive catalyst for this shift. Rather than simply automating work away, AI is redistributing economic value across the skills spectrum. Demand is falling sharply for routine content creation and structured cognitive work, such as basic writing and routine analysis. Meanwhile, demand is rising for skills that AI cannot easily replicate, such as expert judgment, coordination, and regulatory compliance.

The authors describe a “striking and persistent disconnect between what workers choose to signal and what employers truly need to get work done.” The less valuable generalist traits include leadership, communication, teamwork, and problem-solving. On the other hand, workers tend to “under-signal” the hard, specific, valuable skills that employers want—and in many cases, they “lack them altogether.”

According to Bradlow, the rise of large language models is making deep expertise more—not less—valuable, because “who’s going to train it? A person with deep skills. Who’s going to assess whether it’s correct? Someone with deep skills.”​

The $18,000 swing

One of the most revealing findings of the WAsX is that skill value is role-specific, not universal, with a capability that boosts pay in one industry, for instance, actually lowering it in another. For example, the index found that signaling “strategic analysis” skills is correlated with an $8,000 salary increase for sales representatives but a $10,000 salary decrease for technical validation leads. In highly technical roles, the market prioritizes domain depth—such as liposome technologies or catalytic reactor techniques—over general management strategy.

New currency for work

The implication for the future of work is clear: Skills are replacing job titles as the primary currency of the labor market. To navigate this transition, the authors suggest a radical shift in perspective for all stakeholders:

• Employers must move away from “job architecture” and toward “skill architecture,” breaking roles into underlying tasks and aligning compensation with the specific skills that drive value. As Bradlow put it, firms need to understand “the workflow of your job” and then decide “what’s the best role for a human in that workflow.”

• Employees are encouraged to view their careers as a portfolio of high-value skills rather than a list of titles, using AI to rapidly acquire technical depth.

• Educators need to rebalance curricula, shifting away from generalist preparation toward specialized, job-ready capabilities that ensure graduates can contribute from day one.

Bradlow had some advice for people who worry that they don’t have the right skills in the age of AI: Don’t worry about what you lack, and find hard problems to solve. “What you want people to be able to do is to take potentially hard, amorphous problems, structure them, and solve them to the best of their ability.” Whether you’re an English or a physics major or a computer scientist, Bradlow said it doesn’t matter. Your skills come down to the sharpness of your critical thinking.

Bradlow said he gets asked all the time whether he regrets taking so much computer science, since LLMs can do the work now: “It taught me structured thinking. It taught me algorithmic thinking.” The only thing he’s worried about, he added, is the person who doesn’t know how to take hard problems and break them down and solve them. Furthermore, he added, LLMs have democratized skills by allowing people of different backgrounds to ramp up their expertise a lot faster than before. “So it’s not obvious to me that the losers are the humanities people and the winners are the quant techie people.”

“I would say the same to everybody,” Bradlow advocated. “Learn how to take hard problems, break them into small structured problems, state what your assumptions are, understand uncertainty, and you’ll be fine. There are jobs for those people all the time. Those are called thinking and structuring skills.”