This post was originally featured on ScoreNYC

The robot revolution is here, and it’s taken over the manager’s desk — or, at the very least, begun to share the chair.

In recent years, artificial intelligence (AI) tools have begun to edge their way into the day-to-day conventions of work and leadership. In 2019, the Oracle and Future Workplace AI@Work Global Study reported that nearly 50 percent of respondents said that they currently used some form of AI at work — a notable leap from the year before, when only 32 percent said the same. However, more striking are the attitudes that the data revealed; according to the study’s researchers, HR leaders were the most optimistic about AI’s entry into the workplace (36 percent), followed by managers (31 percent).

This revelation might seem counterintuitive at first glance. After all, nuanced interpersonal work doesn’t seem to fall easily into the category of repetitive, data-driven roles that most people associate with AI expansion and automation. Fields such as retail, transportation, advertising, and logistics all seem more at risk for technology-caused job losses. In fact, CNBC recently reported that 42 percent of workers in the business and logistics support sector have “above-average concerns about new technology eliminating their jobs.”

And yet, AI is just as present in the management profession — if, notably, not as feared. The above Oracle study noted that most employees seemed to view AI and human managers as complementary, rather than competing, presences in the workplace.

Respondents shared that they felt robots were better than human supervisors at “hard” organizational skills such as maintaining work schedules (34 percent), problem-solving (29 percent), and managing a budget. However, they also noted that humans surpassed AI when it came to “soft” empathetic skills like understanding employee feelings (45 percent), supporting a positive workplace culture (29 percent), and evaluating performance (26 percent).

These findings suggest that AI is in a position to support human managers, not replace them. Currently, more than 50 percent of a manager’s time is spent on administrative coordination and control — the very tasks that, as the Oracle study indicates, employees believe are better performed by AI. Conventional managers spend just 30 percent of their time on problem-solving and collaborating, 10 percent on strategy and innovation, and a mere 7 percent on developing their people and engaging with stakeholders.

Given this data, it stands to reason that if managers delegate the brunt of their administrative responsibilities to AI tools, they will have more time to spend on the tasks that they are best-suited to perform. Managers already want AI support; according to data published in a 2016 issue of the Harvard Business Review86 percent of surveyed managers mentioned that they would like AI support with their monitoring and reporting responsibilities.

Manager-to-AI delegation stands to benefit businesses, as well. To quote American Express’s vice president of customer data science and platforms, Anthony Mavromatis, in an article on the matter for WeForum: “By cutting [managers] loose from tasks traditionally expected of them, AI allows managers to focus on forging stronger relationships with their teammates and having a greater impact in their roles.”

Layne Thompson, the director of ERP Services for a U.S. Navy IT organization, made a similar point in an article for the Harvard Business Review. He noted, “More often than not, managers think of what they’re doing as requiring judgment, discretion, experience, and the capacity to improvise, as opposed to simply applying rules. And if one of the potential promises of machine learning is the ability to help make decisions, then we should think of technology as being intended to support rather than replace [managers].”

AI has tremendous promise to help managers — and a few potential pitfalls.

Earlier this month, the Verge ran an article about the dark side of AI-supported management, profiling what can happen when human managers become subordinate to AI organizers rather than partners with them. The piece shared a few anecdotes of management gone wrong, particularly highlighting the case of an Amazon warehouse worker who was driven to injury under AI-directed management processes.

“Management was completely automated,” the worker, speaking under the pseudonym of Jake, told reporters. He explained that managers would patrol the warehouse floor with their laptops open, pushing workers to speed up whenever their AI-powered tracking software indicated a slowdown.

The repetitive, fast-paced work took a toll on Jake’s back, eventually creating a disc injury. One manager told him that bending his knees more when lifting would be easier on his back — but when he did so, another supervisor came by to notify him that his rate had dropped and he needed to speed up. Eventually, the problem became so pronounced that Jake was unable to continue working.

Jake’s story illustrates what can happen when humans are used to reinforce AI-produced hard metrics, rather than work in collaboration with AI to support employees. In Amazon’s case, prioritizing the achievement of what AI viewed as optimal performance led to managers giving up an essential aspect of their usual roles — noticing when employees are struggling and intervening when necessary. As a result, Jake was unable to work, and the company had to pay the turnover costs of replacing a good employee.

AI holds tremendous potential for managers, but it must be applied correctly. AI’s role should always be to support managerial performance — not to set performance metrics and use human managers as enforcers.

When businesses incorporate AI management technology into their workflow, they need to have a clear vision for how it will — and will not — be used. Leaders should emphasize that the technology will only be used to automate time-consuming administrative work and give managers more time to fulfill their interpersonal responsibilities, such as providing team-wide support, developing personnel, and pursuing innovation. Then, they should periodically check in to make sure that those goals are being met and that AI hasn’t — as in the case of Amazon’s Jake — turned toxic to the very people it was intended to help.

There is no doubt that AI can benefit managers, employees, and businesses as a whole in the future. However, business leaders will need to be strategic, and, above all else, demonstrate human empathy as they incorporate the technology into their managerial ranks.