With the stress of managing 83 employees taking its toll, Hannu Rauma was feeling discouraged and frustrated.

“I was getting too bogged down in all these things that were going wrong amongst the teams, and feeling this disappointment,” says Mr Rauma, who is based in Vancouver, Canada.

He is a senior manager at a company called Student Marketing Agency, which employs university students to provide marketing support for small businesses.

“When I was bringing new clients on board, half of my mind would be saying, ‘we’re going to screw up’, and it would dampen my enthusiasm.”

But Mr Rauma says that all changed from last November, when the firm began using an autonomous AI manager developed by US-based company Inspira.

The AI manager helps the agency’s employees, who work flexible hours remotely, to set their schedules and plan their workloads in advance.

It checks their timekeeping, sends them deadline reminders and regular check-in messages, and records the time spent on different clients, so the latter can be billed accurately. The AI also makes suggestions to improve the wording of written text, is available to answer work-related questions, and automatically updates everyone’s work progress in a central portal.

Mr Rauma says that the shift towards an AI manager has not only reduced his stress levels, but has enabled his employees to work faster and be more productive. “I’m able to focus on the growth of the company and all the positive things. It’s added years to my life, I’m sure,” he says.

Mr Rauma adds that his relationships with his employees have also improved drastically. “Before, it felt a lot like a father-child situation. Now, we’re more on an equal footing. Before, it was only about solving problems. But now we’re able to have more light-hearted discussions.”

But not everyone at Student Marketing Agency is using the AI manager yet. Mr Rauma and 26 of his 83 employees were actually part of a study run by Inspira and academics from Columbia University, Arizona State University, and the University of Wisconsin to compare the performance of the AI manager with its human counterparts.

Participants were divided into three groups: one coached by a human manager, another by the AI manager, and the last group by both AI and human manager.

The AI manager achieved a 44% success rate in getting employees to pre-plan their workdays in advance, and was able to motivate the employees to log in on time 42% of the time. These figures were comparable to the human manager, who achieved scores of 45% and 44% for those two areas.

Yet when the AI manager worked in partnership with a human manager, together they achieved a 72% success rate in getting employees to pre-plan their workdays, and managed to achieve 46% on-time success.

Despite the study being statistically small, and concentrated on a specific type of worker and field, its results point to interesting implications for companies introducing AI tools.

While businesses like UPS, Klarna, Dell and others have announced significant job cuts this year, with the intention of replacing many roles with AI, Prof Paul Thurman, from Columbia University in New York, argues that swapping management roles completely for AI would be a mistake.

“The middle management layer is the most critical layer in any organisation,” says the professor of management. “They’re the layer that, if it starts turning over, you’re in for a wild ride. Your people don’t see continuity, they don’t get mentoring and coaching… all the human things that human managers are better at than AI and should be focusing on.”

AI, Prof Thurman adds, can liberate managers from endless reminding and checking in, to focus on more innovative ways of working. For example, managers can cherry pick project teams based on individual skillsets, oversee the brief, then hand over to their AI to manage minutiae like deadlines.

AI can also identify who in the team is falling behind and may need to be managed more closely by a human, and by the same token, hone in on star performers who require extra recognition.

But companies should steer away from AI managers becoming a surveillance tool, he says.

“You don’t want to get to a point where you are noting that, not only do people not clock in on time, but they take too much time at lunch, and they’re not eating enough salad. You don’t want to go that far,” says Prof Thurman. “You want to find the right way to encourage the right behaviours.”

AI managers can also help people who have become “accidental managers” – people who excel in their roles and end up managing people as a result, despite management not being a natural skill for them, says Tina Rahman, founder of London-based HR consultancy, HR Habitat.

“We did a study which looked at the reasons people leave a job. Almost 100% of the respondents said it was because of bad management.

“Some of them said they didn’t like the way they’d been managed, and most of them also said it was because they didn’t know what was expected of them or if they were doing a good job,” says Ms Rahman.

“You’d assume that an AI manager would be built to give those correct instructions, to give complete transparency on the requirements, and the outcomes. People are likely to be more productive when they know what’s expected of them.”

But an over-reliance on AI management sets the tone that companies only care about output and not people, Ms Rahman warns.

“It’s going to be very hard for a business to tell their employees that they’re introducing this brand new AI system that’s going to completely manage them, then say, with the same face, that ‘we care about your experiences in the workplace,’” she says.

Yet perhaps the biggest concern about AI managers is not from a people perspective, but from a cybersecurity one, warns James Bore, managing director of cybersecurity consultancy, Bores, and speaker and author.

“If you have an AI manager, and you’ve given them all of the company’s processes, procedures, and intellectual property that is suddenly all in the software, it can be kidnapped by someone who wants to clone it, and it could also be held to ransom,” says Mr Bore.

“If you’ve come to rely on it, which companies will when they start replacing humans with AI, you’re kind of stuck, because you’ve got no resilience, no option to switch back to the humans, because you don’t have them anymore.”

Rather than companies becoming more efficient through an extensive use of AI, Mr Bore says there could be an unintended consequence beyond becoming dependent on systems that could fail.

“The more you automate, and the more you remove people from your business, yes, you’ll bring down costs. But you will also make your company more replaceable.”