Mental Health

‘Burnout tech’ seeks to identify signs of workers’ mental distress by reading Slack messages and email

Two tech companies that use the software say the worker data is totally anonymous.
article cover

Photo Illustration: Dianna “Mick” McDougall, Photo: Getty Images/AnVr

· 6 min read

Whether employees are griping about the background music before a virtual all-hands meeting, or spending their day gleefully sharing the dankest memes over Slack, these casual communications may contain within them the early warning signs of burnout, according to purveyors of new technology intended to monitor worker discontent.

Just as technology exists to monitor workers and identify when they might be slacking off, other software is now being used to flag what an algorithm detects as indicators of burnout, after reading written text on a variety of platforms, such as Microsoft Teams, Slack, and email.

Recognized by the World Health Organization as a “syndrome conceptualized as resulting from chronic workplace stress that has not been successfully managed,” burnout has soared during the pandemic and become entrenched in the cultural zeitgeist as a malady induced by overwork. Burnout has been reported by professionals across a multitude of sectors, and it’s now common for employers to devote resources to address the pervasive issue.

As part of this effort, some employers are turning to software that can be deployed to identify warning signs of burnout before they manifest beyond the silo of a messaging platform. “The technology is still in the early days…so we’ll see more and more of this sort of application of AI in the coming years,” said Quinn Underwood, the CEO of Autumn, which makes burnout detection software, using an algorithm that integrates with Slack and employee calendars.

Autumn allows clients to “begin measuring trends in how you and your team are doing, so you can identify issues before they become fires,” as the company’s website reads. Users—from managers to the employees working beneath them—opt in to the service voluntarily across the 16 organizations currently using the product, Underwood said.

Erudit AI, another maker of burnout detection software with “over two dozen” clients in the US, sees demand from companies grappling with retention issues, according to CEO Alejandro Martinez Agenjo. Companies “don’t have the technology to really understand the drivers which are moving people to quit,” he told HR Brew. In marketing materials reviewed by HR Brew, the company says its technology “helps management better understand employees without having to survey by providing daily metrics such as burnout risk, engagement, and turnover risk. It also alerts executives to issues and the true sentiments of the company and each department.”

Very sensitive. But Pamela Dixon, executive director of the World Privacy Forum, has concerns about the technology’s potential for misuse by employers. She told HR Brew that burnout is “a subjective judgment,” and using this technology in the workplace “is an area of great sensitivity.”

“We have to be very careful that technology isn’t mistaken for a diagnosis,” Dixon said. “And that can be a very problematic situation if it does happen.”

Jon Thurmond, an HR manager with 22 years experience of working in the construction industry, expressed similar reservations: “If I thought my employer was scanning my messages and conversation to figure out if I was unhappy, that is very unnerving to me. They’re doing it today for [burnout], what are they going to do tomorrow? What are they going to do the next day?”

HR is challenging. HR news doesn’t have to be.

HR Brew keeps you effective in the fast-changing business environment.

“Not creepy.” Both Autumn and Erudit AI compile aggregated measurements of burnout in teams and feed the data into a digital dashboard. For Erudit, data is collected from teams of seven or more, while Autumn tracks communication of teams of four workers or greater. Both companies said all data is anonymized and cannot be traced back to any individual worker.

A screenshot of the Erudit AI "burnout dashboard"

Erudit AI

On Erudit AI’s dashboard, above, a manager can “see everything in real time—they can see burnout levels, engagement levels…impacts—which are events that are having a great impact into the company—mood, or whatever they want,” Agenjo explained. Later, he added, “When you take a look [at] the dashboard, you’ll see that this is not a creepy thing.”

Erudit AI’s algorithm was built by a team of psychologists, using the Maslach Burnout Inventory—a clinical diagnostic tool used to gauge work-induced burnout. The psychologists read random messages taken from social media platforms, and trained the AI to judge text according to level of intensity, a March company report explains. Anyone with a manager’s license purchased from the company has access to the dashboard—typically, that means team leaders and executives, but “companies are free to share the information, such as burnout risk, with the rest of their organization should they choose to,” Agenjo wrote in an email.

Autumn’s algorithm learns from a certain feedback loop between users and the AI, Underwood said. It was built by assessing user responses to diagnostic surveys such as the PHQ9, which is used in clinical settings to gauge depression and anxiety. “We built our AI model to essentially identify patterns in language that are reflective of many of those kinds of self-reported assessment scores,” he explained.

When it comes to interventions to address burnout, the onus is on the client—at least for Erudit AI. “We diagnose and we provide to you the context. But we are not providing solutions,” Agenjo explained.

Once a burnout signal is detected, Autumn works to provide “resources down the care continuum,” which can include reminding employees of a company’s existing mental health benefits, or discussions about scheduling a vacation, or providing the opportunity for a mindfulness activity, Underwood explained.

Oh yes, privacy. For Dixon, the advent of burnout tech comes at a time of “rapid and global expansion of datasets in the public-health arena.”

That datasets could potentially lead to a punitive response from an employer were foremost among her concerns: “Can this be used as a reason for firing someone? Can this be used as a factor in giving a person a lesser work detail?”

Both Agenjo and Underwood stressed that their algorithms prioritize user anonymity. “We are not talking about individuals; we’re talking about aggregate teams, over seven people. So that’s not considered health information,” Agenjo said.—SB

Do you work in HR or have information about your HR department we should know? Email [email protected] or DM @SammBlum on Twitter. For completely confidential conversations, ask Sam for his number on Signal.

HR is challenging. HR news doesn’t have to be.

HR Brew keeps you effective in the fast-changing business environment.