Are Relational Skills Becoming More Valuable Than Technical Skills?
A decade ago, computer scientist Geoffrey Hinton, sometimes called the “godfather of AI,” told an audience, “People should stop training radiologists now.”
Hinton’s prediction was based on AI’s increasing ability to analyze complex images, one of a radiologist’s core tasks. His prediction echoed an often-repeated assumption: that automating major parts of a job eliminates the need for the job itself.
But a decade later, the problem facing radiology isn’t that there are too many radiologists without jobs. It’s that there are too few radiologists to meet growing demand.
So, what happened?
Artificial intelligence didn’t eliminate the radiologist’s job. It changed its composition and skill profile. Labor economists Daron Acemoglu and Pascual Restrepo have shown that technology doesn’t simply eliminate work. It changes tasks humans perform, and, in turn, the relative value of the skills required to perform them.
In 2026, radiology professors Frederico Bruno and Alessandra Spelendiani wrote in an article aptly titled “Soft Skills in Radiology: The Human Core of a Technological Specialty” that as AI automates imaging tasks, “uniquely human strengths such as judgment, ethical reasoning, empathy, and leadership will become even more central to radiologists’ value.”
The shift in radiology work reflects a broader change underway in many occupations. Artificial intelligence is changing job compositions and, therefore, which skills matter most.
While there’s been much commentary about the importance of human skills in the age of AI, emerging evidence suggests one skill set may become especially critical: relational skills. Unlike other job-specific human capabilities, nearly every job requires working with other people.
As AI takes on more technical and operational work, relational work may command an increasing premium. The problem is that many organizations aren’t prepared to develop, measure, and reward the skills to do it.
The Relational Premium
Imagine two lawyers at the same firm. Both are experts in their field, but one is especially skilled at analyzing complex documents, and the other excels at building client relationships. Now imagine this firm begins using AI for document analysis tasks.
AI won’t eliminate the need for lawyers, but it will shift what they spend their time doing and, as a result, make relationship-building skills (and the lawyers who have them) more valuable.
Economists Lukas Freund and Lukas Mann recently modeled this dynamic, using data on the tasks people perform across occupations to estimate what happens to workers and wages when AI automates some of their tasks. They project that as AI takes on more information-processing tasks, “the return to social skills rises, that to analytical skills falls.”
Social skills already carried a premium long before generative AI. Harvard economist David Deming and Yale economist Lisa Kahn analyzed millions of job postings from 2010 to 2015 and found that jobs requiring more social skills were associated with higher pay and better firm performance, with evidence that employers especially valued jobs that required both.
This was echoed in a 2026 study by Oxford University labor market researcher Elina Mäkelä and colleagues, who analyzed over 30 million job postings and observed that AI-related roles requiring skills for “working with others” tended to offer higher salaries than AI roles that didn’t.
In the World Economic Forum’s 2025 Future of Jobs survey, the share of employers identifying social influence and leading others as a core skill increased by 22 percent in two years. This was the largest increase of any skill category measured.
This doesn’t mean technical skills are becoming unimportant. But as AI takes on more technical work, relational capability may become a key differentiator.
And if relational skills are becoming more economically consequential, organizations need to stop treating them as noble personality traits or “soft skills” and start treating them with the same rigor as technical skills.
Get Zach Mercurio, PhD’s stories in your inbox
Join Medium for free to get updates from this writer.
The problem is that many organizations don’t appear ready to meet this growing need.
The Relational Skills Gap
In a 2024 poll of over 2,000 employees, just 35 percent said their organizations offer relational skills training, while nearly half reported that managers and team leaders have a major skills gap in communication and interpersonal skills.
On the employer side, a National Association of Colleges and Employers’ 2025 survey showed 96% of employers rated communication as highly important for new graduates, but only 54% rated recent graduates as highly proficient in it.
In my experience working with leaders across diverse types of organizations, this isn’t an awareness problem. Most leaders and organizations know relational skills are important. It’s a rigor problem.
We’ve collectively called relational capabilities “soft.” Any time we see something as “soft” or “simple,” we tend to approach it with less intention and rigor. As a result, much of the existing “soft skills” training is optional or looks like a one-off workshop, leadership development series, or a few online modules.
Most organizations would never treat technical skills that way. For example, consider training someone on the technical skill of financial modeling. We’d break the capability down into specific, demonstrable skills like constructing formulas, building models, and analyzing scenarios. Then we’d give real problems to solve to practice those skills, evaluate the quality of their work using agreed-upon criteria, and require demonstrated proficiency to occupy a role, get promoted, or get paid more.
Now consider how we tend to train someone to do something like give developmental feedback. We send them to a workshop, teach a framework, or have them go through a required supervisory course. But it often stops there.
How often do we break relational capabilities down into discrete skills, give people real scenarios to practice those skills, observe and provide feedback, and require them to demonstrate proficiency before occupying, say, a leadership role? What are the consequences for not doing it well or the rewards for proficiency?
The problem isn’t that people simply lack relational skills. It’s that many organizations lack rigorous systems for developing them.
Treat Relational Skills Like Technical Skills
To solve this problem, organizations should apply the same rigor to relational skill-building as they do to technical skill-building by defining, developing, measuring, and rewarding them.
1. Define. Start by developing more specific, operational definitions of the relational skills required, rather than using broad labels like “people skills,” “communication,” or “emotional intelligence.” For example, “active listening” is a category of skills, not a skill itself. “Before responding, accurately summarize the other person’s perspective and check for understanding” is an observable skill. Relational skills can usually be grouped into four major categories: skills for understanding, skills for connecting, skills for communicating, and skills for navigating relationships.
2. Develop. Knowing how to do something isn’t the same as being able to do it. Relational skills require observed practice to hone. Give people real opportunities to practice specific skills, evaluate their performance against set criteria, provide feedback, and repeat. The Cleveland Clinic, for example, trains clinicians in this way by operationalizing specific relationship-centered communication skills and having clinicians practice them in realistic scenarios with observation and feedback. In a study of more than 1,500 physicians, patient communication ratings, perceived empathy, and burnout improved significantly.
3. Measure. Measure proficiency in relational skills regularly. Move beyond assessment instruments that measure employee sentiment, such as engagement surveys, and start measuring people’s experiences with each other’s observed relational skills. For example, instead of determining if someone is a “good communicator,” assess the frequency with which others observe them demonstrating the skills that meet the criteria of good communication through direct observation and evaluations.
4. Reward. Finally, demonstrated proficiency in relational skills should have clear consequences. If someone wouldn’t qualify for a technical role without demonstrating the technical skills to do it, they shouldn’t qualify for a leadership role without demonstrating the relational skills to do it. Tie promotions and pay to relational proficiency, along with technical proficiency. This shift in rewards is already occurring. In August 2026, Ernst & Young announced a $100 million investment to recognize skills and behaviors it believes will differentiate the firm in the future, including collaborating with and leading others. As its chief talent officer put it, “How we reward our people defines what we value as a firm.”
Rethinking What It Means to Be Skilled
It’s been over a decade since Hinton predicted we should stop training radiologists, but we still need radiologists. What’s changed is what it means to be a skilled radiologist.
As AI takes on more technical work and is deployed in more organizations, the ability to complement that technology and understand, connect with, communicate with, and navigate relationships with other people will only become a more important part of the job.
The question is whether our organizations, and the people who comprise them, are prepared to capitalize on that shift.
Zach Mercurio, Ph.D., researches purposeful leadership, meaningful work, and mattering, and wrote The Power of Mattering: How Leaders Can Create a Culture of Significance (Harvard Business Review Press). He helps leaders and organizations create cultures where everyone feels noticed, affirmed, and needed.
Explore more of his research, writing, and resources at zachmercurio.com.
