It’s become increasingly clear that AI is going to shape workplaces in ways that we can scarcely anticipate.

Still, when we think about ensuring that the output of any human–machine collaboration is accurate, it’s easy to assume that humans will be the ones double-checking machines, with the latter’s well-established propensity to hallucinate. But we humans have our own biases, too. So what happens when it’s the machines looking over our shoulders?

This week, we’ll discuss. Plus, how should leaders conceive of a workplace with not three or four or even five but six generations of employees?

Umpiring the umpires

Today, many high-stakes fields, including finance and air-traffic control, are adopting AI oversight systems to improve accuracy. But how does this change the behavior of the people whose behavior is suddenly under the microscope? And are these changes all positive?

For clues, it’s instructive to look at professional athletics, which has long relied on the technology to ensure its biggest calls are made correctly.

In 2006, for instance, the Hawk-Eye system was introduced into competitive tennis. Hawk-Eye uses computer-linked cameras around the court to create a 3D representation of the ball with respect to the service lines, and it’s very accurate, performing with an average error of 3.6 millimeters. Hawk-Eye is then used to check the work of human umpires, with players being allowed a few challenges each set.

New research by Kellogg associate professor Daniel Martin and Kellogg PhD student David Almog uses a robust set of Hawk-Eye data to analyze the technology’s effect on how umpires call matches. What they find: in general, umpires’ accuracy improves when they know that AI could contradict them; their overall mistake rate dropped by 8 percent after the introduction of Hawk-Eye.

Which sounds great, right? Curiously, however, there were certain situations in which umpires’ mistake rate went up after AI came on the scene. Close serves were the most noticeable: the mistake rate for those rose by 22.9 percent. And the sudden rash of errors, the researchers found, stemmed from a newfound tendency for umpires to call balls in that were actually out.

The researchers believe this shift stems from a desire to avoid a particularly thorny situation: if an umpire stops a point by calling a ball out when Hawk-Eye shows it was actually in, there’s no easy way to fix the error. After all, you can’t put the ball back into motion and resume play where it left off. Instead, the umpire has to decide whether to replay the point or award it to the challenger. It doesn’t help that the umpires learn of their mistakes in a very public way: the AI analysis of the ball and the court line are projected on a large screen, with players and fans reacting—stirring up feelings of shame, pride, or embarrassment.

To be clear, the researchers find that the combination of umpires and AI led to more accurate calls across the board. But the fact that it changed umpires’ behaviors in unanticipated, and not entirely positive, ways is important, say the researchers, especially as its use spreads to other contexts.

Imagine, for example, a pretrial bail judge decides to allow a defendant to go free—only to be overturned by an algorithm that determines the person is too dangerous to be let out of jail. The threat of being overruled might shift the judge’s decisions, says Almog, “so they might be biased to release fewer people.”

You can read about their research in more detail here in Kellogg Insight.

Leading across generations

Are you prepared for 6G? No, that’s not a new tech standard for cellular networks. It’s the six- generation workforce. Because with Silent Generation octogenarians still working, and the oldest members of generation Alpha teenagers just starting to apply for summer jobs, organizations are now confronting, and accommodating, a more age-diverse workforce than ever before.

To do so well won’t be easy, says Kellogg clinical associate professor Nicholas Pearce, writing in Harvard Business Review. He points to increasing congestion in the talent pipeline as a particular concern:

At one end is often a bottleneck of more seasoned workers in top positions who, as they approach “retirement age,” are either financially unable or psychologically unwilling to retire and have nowhere else in the talent pipeline to advance. At the other end is often younger talent impatiently waiting their turn for advancement into more challenging roles because the pipeline is clogged. In the middle are the so-called “sandwich generations,” sitting frustrated as the unwritten rules of the game change right before their eyes. While these challenges are not entirely new, the 6G workplace exacerbates them.

Pearce suggests creating meaningful opportunities for senior talent to continue to bring value to the organization outside of executive roles, perhaps by training or mentoring younger employees. He also suggests putting increased emphasis on the things everyone wants from work, like purpose, meaning, and growth.

You can read more from Pearce in HBR here (paywalled).