“A concerning picture” | Working with AI makes employees lonely & less healthy, study finds

Working with artificial intelligence (AI) could make employees lonelier and less healthy, a Harvard Business Review study has found.

The study, authored by researchers David De Cremer and Joel Koopman, finds that while AI can deliver major productivity gains for organizations, this may be at the expense of workplace relationships and social connections.

De Cremer and Koopman state they aimed to investigate how working with AI affects the connection employees have with human coworkers.

The researchers conducted a series of studies, examining different types of AI in multiple global markets and varying industries including a Taiwanese biomedical company, an Indonesian property management company, and a Malaysian technology company.

“A concerning picture for the well-being of employees using AI”

In the first study, De Cremer and Koopman interviewed 166 engineers from the Taiwanese biomedical company, who had worked with the company for an average of three years and with AI systems for two years.

The results showed those who worked with AI reported “greater feelings of loneliness,” with knock-on effects including greater alcohol consumption and insomnia.

Looking for more

Sneaky GPT | More than half of workers are using AI, but does it really matter?

The researchers then sought to verify the correlation through several further studies, including one group of 120 Indonesian real estate consultants, and another group of 294 employees across operations, accounting, marketing, and finance at the Malaysian tech company.

Each employee in the studies had on average been using AI as a daily tool to support their work for over 18 months. Both companies assigned some employees at random to pause their use of AI for three days.

De Cremer and Koopman say those who continued to work with AI had a greater desire for connection and were more lonely, again with possible knock-on effects including greater alcohol consumption and insomnia.

“Organizations need a fundamental shift in mindset”

The authors recognize the potential of AI in the workplace, noting its uses from rapid analysis of data or legal documents to sales forecasting and candidate screening.

The possibility of huge gains in productivity have understandably led to widespread adoption.

IBM figures from 2024 put the percentage of enterprise companies (over 1,000 employees) already using AI at 42%.

According to MIT Sloan, that share grows to 50% for companies with over 5,000 employees, and to 60% for companies with over 10,000 employees – though for small businesses, it drops to just below 25%, according to the US Chamber of Commerce.

Read more from us

'This does a better job' | Morgan Stanley hopes its new AI assistant will save its wealth advisors thousands of hours of labor

Morgan Stanley, for example, believes its new AI assistant will save its 15,000 advisors over 500,000 hours of labor per year by automating routine tasks. Mastercard says AI is helping it to schedule candidate interviews 90% faster, and Microsoft claims its AI-powered HR Virtual Agent bot has saved 160,000 hours for HR service advisors by answering routine questions.

But De Cremer and Koopman argue that to avoid causing isolation, loneliness, and disengagement in workers who use AI, employers need a “fundamental shift in mindset.”

“Instead of viewing AI merely as a means to automate and optimize, they must see it as a tool for enhancing the human experience at work,” they write, adding that AI systems should be used to enrich rather than replace employees’ jobs.

“For example, AI can take over more tasks, and at the same time leaders must create dedicated spaces and times for employees to connect face-to-face,” De Cremer and Koopman continue. “This might mean carving out time for team-building activities, social events, or even just casual coffee chats. The goal should be to foster a culture where social interaction is valued and encouraged, not seen as a distraction from “real work.”

“Focus on humans first and AI second”

The study makes several recommendations to employers and, by extension, HR teams.

Namely, De Cremer and Koopman suggest companies should monitor employee wellbeing and social embeddedness, citing employee engagement, job satisfaction, and perceived social support as three valuable metrics to track.

They also propose redesigning workflows around the strengths of humans and machines, rather than layering AI on top of existing processes. This may include allowing employees autonomy and control over how and when they use AI.

By doing so, the researchers argue companies can mitigate human costs such as reduced job satisfaction, motivation, and wellbeing.

“If you want to make AI adoption projects successful and viable, you need to focus on humans first and AI second,” De Cremer and Koopman state.