Skip to main content
AI and machine learning January 23, 2023

HR teams can (and should) use AI to predict employee burnout

Frustrated african american woman keeping eyes closed and massaging nose
In the fall of 2019, Wenche Fredriksen took to LinkedIn to share a personal story of burnout and recovery.

At the time of the burnout, she was 37 years old, the mother of two young children, and a manager at a consulting company. From the outside, she wrote, it appeared ideal. Burnout made for a different reality: “I had no energy. I could not handle sound, light, or movement. I could not focus or handle information. I could not eat, sleep, or cry. My hard disk was full, and my batteries were flat. I felt like a total failure. I just wanted to disappear, sleep, and have no one expect anything of me, ever. I was convinced that I would never work again.”

Across populations, geographies, and sectors, and whether on the factory floor or in the C-suite, employee burnout is real and devastating. The symptoms are common – emotional and physical exhaustion leaving them unable to focus on themselves, their work, or their loved ones.

These are not rare and isolated cases. In Europe, some 1 in 10 employees are affected by burnout. According to one report, worker burnout in Germany is said to cost its economy some €9 billion annually in lost productivity. US figures are even grimmer – estimates are that US employers lose over $300 billion each year. Global acknowledgment came with the World Health Organization’s decision in 2019 to include burnout as an “occupational phenomenon” in the 11th Revision of the International Classification of Diseases. And that was prior to the crises of the coronavirus pandemic, during which burnout – especially among professionals in the medical sector – became a household word.

While the reported cases and their impacts are sobering – especially in these last few years – surprising developments in artificial intelligence offer new hope for identifying and combatting the stress that leads to burnout. What, then, are these opportunities, and how can business leaders use these tools – sooner rather than later – to change their workplaces for the better?

Assessing the risk of burnout and intervening on behalf of personnel remains the responsibility of trained HR professionals. We need their skills and good judgment in handling what are often complex and highly sensitive situations.

However, artificial intelligence can be useful in identifying who is at a higher risk of experiencing severe burnout. This information can aid busy HR teams in prioritizing specific teams, offering resources, and monitoring developments for additional intervention.

But how exactly can AI help us in this way?

Our organizations already produce a wealth of electronic trace data about how we work. Email, for example, is widely used (and overused). It is a readily accessible data source of collaborative behavior and, as many acknowledge, a source of strain and stress itself. A 2015 technology research firm Radicati report claimed that workers received an average of 121 email messages daily. In the Adobe 2018 Consumer Email Survey, US white-collar workers reported that they spend an average of 3.1 hours checking work email each weekday.

In a study of a medium-sized organization (read the research), my fellow researchers and I applied prediction models to the company’s email traffic to predict which employees were at risk of burnout. We confirmed our predictions using survey data and reach an accuracy of over 80 percent.

Such studies are just the beginning – a promising development that deserves replication in diverse and larger companies. We must show whether the results can be replicated, which of the email indicators are important across organizations, and how far in advance the predictive models’ flags should be raised. But, of course, monitoring employee email communications raises questions that must be addressed by correct handling. That is, we must ensure that systems designed for employees address workplace surveillance concerns by prioritizing employee data privacy, even if identifiable information such as sender, recipient, and time are required to predict burnout risk and mobilize HR resources.

As remote and virtual collaboration is normalized, the separation of work and life spheres becomes increasingly blurred. Wenche Fredriksen’s personal story of burnout is far from unusual and may be increasingly so. Organizations can and should do more to prevent such experiences from occurring. The opportunity for HR teams to proactively handle extreme employee stress and burnout can be significantly improved by AI. Reducing the cost to personnel and company bottom lines is worth the effort.

Add new comment