As firms eagerly scramble to integrate AI into their workflow, it can feel a bit like throwing spaghetti at the wall. Maybe AI will improve customer service? Maybe it can focus on low-level work? Maybe it will help with analysis, streamline supply chains, or write better executive talking points?
This experimentation to see what sticks can end up being frenetic and unproductive—and can sometimes go very badly. Just ask the lawyers who have been fined or sanctioned for writing briefs that cite cases that AI hallucinated.
Richard Jolly, a clinical associate professor of Management and Organizations at the Kellogg School, says there’s a better way to decide which jobs should be done by people and which can be more-efficiently handled by AI. Before parceling out tasks, he urges leaders to think through a key question: “Is this one of the things whose value is that a human does it?”
The main thing that humans do better than AI is establish and maintain authentic relationships with other humans. “AI might be the end of coding. It might be the end of a lot of things,” Jolly says. “But it’s not the end of relationships.”
Jolly has developed a new practitioner framework—which he describes using the acronym HUMAN—to identify five areas of relationship-based work that AI will not change. The HUMAN framework offers leaders a handy mnemonic for which responsibilities to retain—and important questions to ask before handing off the job to AI.
Insight in your inbox
Receive our newsletters to keep up with the latest research and ideas from faculty at the Kellogg School of Management.
https://insight.kellogg.northwestern.edu/newsletter/insightful-leader
Humanizing
Does this situation require a human touch?
Chatbots can offer a facsimile of empathy—if you’re having a bad day, an app can tell you that sounds rough or maybe coach you to explore your feelings. But AI isn’t in the business of really grasping what drives a person, what they are hoping to achieve, and what solutions they’re seeking, all of which are at the heart of professional relationships.
In a law firm, for example, AI can be very helpful in quickly writing early drafts of documents. But it’s still not a replacement for the human trait of empathy. The law partners who can connect with clients authentically are the firm rainmakers.
“It’s the ability to show, ‘I am intensely interested in you. I care about you. I really want to understand what matters to you,’” Jolly says.
Unscripted
Will human judgment and improvisation be important?
While all these traits contribute in different ways to distinctly human interactions, one trait that is at the core of the kinds of experiences and activities that AI can’t replicate is the ability to be spontaneous and veer off-script when necessary.
People can display vision, pivot when challenged, lean on the trust they’ve engendered from past encounters, and simply connect with their fellow humans at everything from a client dinner to an all-hands meeting.
“People have the ability to walk into a room and get other people excited,” Jolly says. “It’s the ability to stay relational when reality departs from the script.”
Moral
Who owns this?
In 1979, an IBM training manual laid down a rule that has become ever-more relevant nearly a half-century later: “A computer can never be held accountable, therefore a computer must never make a management decision.”
AI can inform all sorts of decisions—from which product to launch to which person to date. But leaders need to ask themselves a deeper, moral question about AI: Is this a decision whose consequences a human being needs to own? Because as much as AI can inform the decision, it can’t take responsibility for it, explain itself to the people affected, live with the consequences, or make amends if it gets it wrong.
“You can never say, ‘Hey, you gave me bad advice. I’m suing you,’ because there’s nothing there to sue. There’s no accountability. And that is a very profound point,” Jolly says.
Humans can build and demonstrate trust in a way that AI can’t. So jobs that require accountability should go to people rather than machines.
Adaptive
Does success require adjusting your priors?
Purposeful disagreement is a crucial part of a well-functioning organization. It makes ideas better. And AI is bad at disagreement.
Left to its defaults, AI makes agreement and confirmation remarkably easy, Jolly says; unless you deliberately ask it to challenge you, it will tend to affirm your thinking rather than test it. That is a real leadership risk. Good leaders need the opposite: exposure to information and perspectives that force them to update their priors when reality pushes back.
“If you are going to make good decisions, you need to consider viewpoints that challenge your own,” Jolly says. “This allows for workshopping and recalibrating your ideas.”
Narrative
Will a compelling story make a difference?
The best presentation deck with the most compelling charts and tables isn’t going to change hearts and minds on its own, Jolly says. The key is the person delivering the information. Persuasive leaders can provide the big-picture vision and customized context needed to get people onboard—something AI cannot do.
Jolly uses the story of a 19th century Hungarian doctor Ignaz Semmelweis, who compiled extraordinary evidence that maternal mortality rates could be cut drastically if doctors washed their hands before delivering babies. His data suggested that particles from the dissecting room were being carried into the delivery room, so he introduced chlorinated handwashing and tracked that maternal mortality dropped from 10 percent to around 1-2 percent through this one intervention.
Though he had the falling death rate as evidence of handwashing’s effectiveness, he couldn’t turn that into an explanation the medical establishment would accept and implement. His failure was not one of science but of creating a persuasive narrative that would change doctors’ habits.
It was only after Louis Pasteur demonstrated how germs cause disease that doctors made it a habit to wash their hands between procedures.
“Semmelweis had the data. He had the mortality counts, the mechanism, the falling death rate in his own ward. But that data changed almost nothing,” Jolly says. “Great leaders build a ladder between the how and the why.”
Keeping the “fun” work for humans pays off for everyone
Following this HUMAN framework isn’t just about making sure you’re using AI in the right way. Thoughtfully delegating work to AI versus humans also pays off in employee satisfaction and company culture.
“It’s an opportunity to make sure that you are having way more fun doing the things you love and are good at—that are also adding value,” Jolly says.
AI can save people a lot of time by drafting routine emails or documents or presentations. That extra time can then go into the relationship-building that constitutes our human superpower. And, Jolly argues, that’s the fun part of a job.
“That’s really where the value added is,” he says. No one is going home pumped up about the awesome email they wrote, but they might feel that way about a powerful interaction they had with a coworker or client.
And keeping yourself and your employees engaged is more crucial than ever. Gen Z has shown disinterest with the modern workplace because of a desire to avoid the mindless meetings and paper-pushing that feels like drudgery.
“They don’t want to feel like they’re spinning their wheels,” Jolly says. On the other hand, AI is perfectly happy to spin its wheels.
“Employees want to feel like they’re making a difference, doing something that’s meaningful, doing the cooler stuff. I think that’s the challenge for all of us: to focus our time and energy on the things where we’re really adding value.”
Kellogg School of Management
Please visit the firm link to site