Work becomes more human in the age of AI

Artificial intelligence is increasingly framed in stark extremes: either as a revolutionary tool poised to transform society for the better or as a looming threat ready to replace human workers. Headlines about “robot nurses,” rogue chatbots, and indefatigable virtual assistants can make the technology feel both exciting and deeply unsettling.

But according to University of Rochester experts from business, medicine, and ethics, the reality is more nuanced. Daniel Keating ’05S (MBA), Kathleen Fear, and Jonathan Herrington say the more important question may not be whether AI will replace humans—but which parts of human work should never have consumed so much human attention in the first place.

Keating, a clinical associate professor of information systems and AI at URochester’s Simon Business School, sees AI less as a wholesale replacement for workers than as an agent that will fundamentally reshape how people spend their time and energy at work.

The technology excels at repetitive cognitive tasks, but human judgment, leadership, and creativity remain essential.

“You have to use it as a creative palette to which you bring your best ideas and hone your strongest questions,” Keating says. “The directive is to have AI make your ideas better.”

More than a machine

“There’s so much hype about AI in the world,” says Fear, senior director of digital health and AI at University of Rochester Medicine. “AI can feel like something happening to people—something they’re not in control of, that will change their lives and jobs without them knowing what to do about it.”

At the same time, the experts see a future in which AI is most useful not when it imitates humanity, but when it creates more room for distinctly human capacities: empathy, creativity, judgment, and connection. And even some of the messier human elements—like mistakes, emotions, intuition, and uncertainty—remain features instead of bugs.

“AI can feel like something happening to people—something they’re not in control of, that will change their lives and jobs without them knowing what to do about it.”

According to Herington, an assistant professor of health humanities and bioethics in the School of Medicine and Dentistry, public conversations about AI often oversimplify the essence of human intelligence.

“It’s just a mistake to think AI will be able to replace all the different cognitive tasks humans are good at,” says Herington, who has a joint appointment as an assistant professor in the Department of Philosophy.

While AI can increasingly perform creative tasks once considered uniquely human, Herington says some dimensions of human thinking remain difficult to replicate.

“There’s a certain degree of taste and style—developing the right questions, selecting the most salient sources, even finding the perfect metaphor—that still matters,” he says.

In Keating’s courses, he explores the current opportunities to drive AI initiatives that matter in the workplace. “AI can make decisions, stitch correlations together, and find novel connections,” Keating says. “It is not a tool; it’s an agent. A tool only does what a human designs it to do. AI is not that: It can do things we did not intend.”

He encourages students to treat AI less as a final authority than as a collaborative brainstorming partner, one that is capable of surfacing unexpected connections and alternative approaches. Keating sees AI becoming deeply integrated into everyday workplace functions—from marketing and financial operations to customer service, project management, and even higher education admissions. Used thoughtfully, he says, the technology can help organizations generate ideas, identify patterns, analyze large swaths of data, and reduce repetitive work. This integration grants the human workforce more time for creativity, judgment, and human connection.

Less ‘pajama time,’ more patient ties

Despite its promise, AI cannot contend with humans in two distinct areas: emotional labor and human discernment.

“When I think about places of shortages—childcare, elder caregivers, nurses, physicians, teachers—human empathy is not yet an ample commodity,” Herington says. “It’s actually really scarce.”

The irony, he argues, is that many modern workplaces force people to spend enormous amounts of time on administrative and cognitive tasks that undermine opportunities for meaningful human interaction.

“When I think about places of shortages—childcare, elder caregivers, nurses, physicians, teachers—human empathy is not yet an ample commodity.”

“Lots of us are doing work that forces us to act in ways where we don’t have time to engage in human empathy because we have to engage in all these cognitive tasks,” he says. “The positive vision for AI is that those kinds of tasks become less important aspects of our job, and we spend more time connecting with other human beings.”

Fear sees AI applications such as Dragon Ambient eXperience (DAX), which records appointments with patient consent, as tools that reduce rote administrative burdens, create more space for human interaction, and mitigate the risk of burnout from too much “pajama time,” a term describing unpaid administrative work that physicians and other healthcare professionals tackle after work hours.

Embrace the mess

At the same time, all three experts caution against treating AI as inherently neutral or infallible.

“As a healthcare institution, we have a serious responsibility to be good stewards of AI, data, and information our patients provide us,” Fear says.

One major concern involves bias and system performance over time. AI tools trained under one set of conditions may behave differently as populations, workflows, or environments change.

“The biggest risk is how tools perform when conditions change over time,” Fear says. “We need to be watching.”

Herington sees another danger in the social dynamics of AI systems themselves. Because many AI platforms are designed to maximize engagement, he worries they may encourage affirmation and dependency rather than honesty or accountability.

“These systems are built to maximize engagement,” he says. “They are almost always going to be sycophantic because that’s what draws people in.”

For him, meaningful human relationships involve friction, independence, and moral challenge. “If you’re in a relationship with someone who doesn’t have their own interests or ambitions, it’s ultimately very unsatisfying,” he says. “What you want is to be with another who has messy ambitions and interests.”

That applies to education, research, and medicine as well.

“It’s never going to speak truth to power,” Herington says of AI systems. “And that’s what you want from good teachers, supervisors, and doctors.

Engagement over anxiety

Rather than downplaying or avoiding AI altogether, Fear believes the best response is informed engagement.

“The most valuable thing people can be doing with AI is getting in and playing,” she says.
Part of her work involves helping clinicians, staff, and patients better understand both the opportunities and limitations of the technology.

“If you understand something, you can shape it,” she says. “The more you know, the more in control you are; the better you can sort out the hype from reality.”

Fear also sees AI lowering barriers that have historically prevented many good ideas from gaining traction. “I’ve been working in innovation for 10 years now, and we have such a smart, creative, passionate community of people here,” she says. “The stumbling blocks often have less to do with the quality of the idea and more to do with resources or technical skill.”

Generative AI tools such as ChatGPT, GitHub Copilot, and Claude could democratize innovation by allowing more people to prototype ideas, build software tools, or streamline workflows without needing coding expertise.

“The sustainable future of AI at URochester Medicine depends on developing a culture where everyone is educated on AI,” Fear says.

‘Junk code’ or genius?

While experts increasingly debate whether AI behaves more like a tool or an autonomous agent, all three researchers agree that human judgment remains central. The future of artificial intelligence is not about replacing humans. It is about redefining what human work should look like in a world increasingly shaped by machines.

If AI can absorb or mitigate repetitive documentation, technical bottlenecks, and administrative overload, they argue, it may allow people to focus more fully on the parts of work machines still struggle to replicate.

“We have just a fraction of the information that AI does, and yet we are able to make these massive leaps as a species with just a little bit of information and all that strange human ‘junk code,’” Keating says, referencing a phrase used by writer Eve Poole.

He points to “the human mess” of qualities such as uncertainty, free will, storytelling, and mistakes as things purposely excluded from AI systems but central to human achievement. These traits, he says, are what make breakthroughs possible: “We write the works of Shakespeare. We create the art of Picasso. We cure polio. In Rochester, we made extraordinary social change through the work of Susan B. Anthony and Frederick Douglass. Every once in a while, we learn to take flight.”

He adds, “All of these seeming ‘deficiencies’ are actually what allow us, as humans, to make those incredible leaps.”