AI researchers face higher risk of loneliness, insomnia and drinking: research

Chatbots are getting better at talking — but that hasn’t saved the workers developing them from an epidemic of loneliness.

According to research published Monday by the American Psychological Association, artificial intelligence researchers are risking their mental and emotional health and are more likely to suffer from insomnia and to drink more after work.

“Humans are social animals, and isolating work with AI systems may have damaging spillover effects into employees’ personal lives,” lead author Pok Man Tang said in a statement.

The study — which surveyed workers across industries and in four countries — examines the impacts on workers of an ongoing “Fourth Industrial Revolution,” spawned by AI.

The study’s relatively small sample size— based on surveys of about 800 office workers around the world — makes its findings suggestive rather than conclusive.

But it suggests early warning signs in the industries that use AI — as well as tactics that managers can use to mitigate them.

The researchers wrote that AI-dependent workers have increasingly found themselves in a very different social environment from which humans are adapted. 

“Across millennia, people evolved internal systems to gauge the quality of relationships with others. These systems have remained effective in a workplace that, just as in primitive tribal communities, prioritized social interactions with coworkers,” the authors wrote in the paper. 

But now, they argue, “the advent of digital, asocial AI systems” is forcing people into ersatz social relationships with robot helpers.

That is causing “a shift toward more of an ‘asocial system,’ wherein people may feel socially disconnected at work,” the authors added.

Such changes are already impacting tech workers in the U.S. and Malaysia, biomedical researchers from Taiwan and real estate consultants in Indonesia. 

In the study, workers surveyed by Tang’s team were assigned to keep working with AI-related technologies or take a three-week break from using them — while they and their families reported the impacts.

The more frequently employees interacted with AI, the more likely they were to respond in both “adaptive” and “maladaptive” ways, Tang explained to The Hill.

Working with AI triggered “a stronger need to socially connect with other human coworkers,” researchers found.  Workers who used AI were also more likely to go out of their way to offer help to other workers — likely because they were lonely and needed social contact, the researchers hypothesized.

But the uptick in loneliness that led AI workers to be more helpful came alongside a rise in other problematic behavior. 

“The same experience that led to greater levels of helping also led to greater levels of alcohol consumption and insomnia after work (which might jeopardize employees’ mental well-being and result in a negative spiral),” the scientists wrote.

One solution is for managers “to try and combat the potential for employees to be lonely,” they wrote. 

For instance, managers should be wary about having too much “density” of AI systems in the office — too high of a ratio of AI harms the ability of employees to “maintain desirable levels of social interactions with others as well.”