'AI is my best friend in my work'

Most employees are turning to AI for far more than task completion — using it for career advice, emotional venting and even companionship — but the technology is doing little to address a loneliness epidemic at work, and may be making it worse, according to new research.

The study, conducted by Constance Noonan Hadley of the Institute for Life at Work at Boston University and Sarah Wright of the University of Canterbury, surveyed 1,545 U.S. knowledge workers who used AI regularly.

It found that 74 per cent were using AI for at least one form of personal or social support.

"The elevation of that level was a surprise to us,” says Hadley, in talking to Canadian HR Reporter.

However, more than half of participants — 52 per cent — still reported feeling highly or moderately lonely at work. Only 12 per cent said AI use made them feel less lonely.

What workers are using AI for

The study, published in the May–June 2026 issue of Harvard Business Review (HBR), adapted the Relationship Functions Inventory — a tool designed to measure how much human colleagues provide career help, personal growth support, friendship and emotional support.

Researchers found that:

• 64 per cent of the workers used AI for career development advice
• 54 per cent for personal growth and skill-building
• 50 per cent for friendship and enjoyment
• 35 per cent for emotional support and coping with stress.

The article quotes one HR employee as saying, "AI is my best friend in my work," and a manager saying AI "makes me feel heard and important."

The versatility of the technology partly explains the rapid adoption, says Hadley, an organizational psychologist and research associate professor in the Management and Organizations Department at the Boston University Questrom School of Business

"It can be shockingly good at helping you figure out how to deal with a conflict, or when you're feeling unsure and unprepared, it can give you a morale boost," she says. "It's really quite amazingly versatile."

While Hadley doesn’t believe the AI companies have necessarily designed AI to become like a work buddy, the nature of the tools automatically lends itself to that kind of interaction.

“It's a learning machine. So, it learns from the people's interaction: ‘Oh, you like it when I tell you a joke? OK, I'll tell you more jokes.’”

The tendency to anthropomorphize AI tools intensified those attachments: 78 per cent of participants used polite terms like "please" and "thank you" when interacting with AI, and 28 per cent described AI using humanlike analogies such as "teammate," "friend" or "personal assistant" rather than terms like "tool" or "platform,” according to the HBR article.

"Many people have a concept in their head of AI as something much more human-like than machine-like," says Hadley.

Why the trend poses risks

The researchers identified four mechanisms through which growing AI use could damage the social fabric of organizations. These include:

• AI depopulating the workplace as tasks shift away from collaboration
• social skills atrophying as workers find AI easier to interact with than colleagues
• the erosion of trust that comes from no longer needing to ask coworkers for help
• a deeper existential loneliness triggered by what one study participant called a "false friendship."

“There’s a lot of reasons to believe it could have a negative effect if it’s not managed.

"It's remarkable, even in my own work, how many times when I would have reached out to a colleague or an expert in the field or a research assistant — I don't do that anymore," says Hadley. "I just don't need to talk to those people anymore."

The productivity gain is real, she acknowledges — but the long-term social cost concerns her.

"Everybody's only talking to their bots and not to each other. I think that is a very bad situation in the long run," says Hadley. "Our data does not indicate that talking with AI makes people more lonely in general, but it also doesn't solve for loneliness. So, we'll have potentially a strong bond and great help from AI and yet increasing disconnection from the other humans."

She adds that she doesn't believe human nature will adapt quickly enough to make AI relationships a substitute for human ones: "I think there are at least going to be many generations to come where loneliness needs to be solved by communications and connections as human beings."

Business case for paying attention

The study found that highly lonely employees reported 27 per cent lower job satisfaction and a 90 per cent greater intention to quit than those low in loneliness, according to the HBR article. Moderately lonely workers reported 13 per cent lower job satisfaction and a 46 per cent greater intention to quit.

Hadley notes the costs go further.

"There have been calculations done by the British government, by the company Cigna, that have estimated the value impact of loneliness on corporations," she says, citing productivity loss, absenteeism, turnover and health costs as key components.

"Medical studies have shown that chronic loneliness is tied to earlier death, higher rates of cancer, dementia, diabetes, cardiovascular issues and so on. The social determinants of health are really starting to come into light."

Hadley also flags a finding that HR professionals may find useful: lonely employees were significantly more pessimistic about AI — more likely to feel their managers were handling AI integration poorly and more likely to fear their jobs would be worsened or replaced.

"Employee social experiences are most likely influencing their behaviours and attitudes toward technological change," the HBR article states. "Paying attention to worker loneliness could thus lead to more-successful AI initiatives."

What HR should do

Despite the growing evidence of AI's social impact, only 33 per cent of participants had received any leadership guidance on how AI might affect their work relationships, according to the article.

Hadley says HR has a central role to play here.

"The first thing I would say is for HR to clarify the priorities of the organization," she says. "If [connections are] a priority then I think it cascades naturally from there to say, ‘Well, then then the next most important thing is to prioritize it in action by investing in data collection,’ so adjusting your pulse surveys and things to capture this possibility of reduced connection – and then investing in ways to support that.”

Manager training is another lever. She notes that many managers may not be aware of how their day-to-day decisions around AI are affecting team cohesion.

"I can imagine they used to do stand-up meetings every day and now they're having one big meeting a month," says Hadley, citing the importance of “being aware of those decisions and how they might affect the relationships that people have.”

Using AI to boost interactions

But her most practical recommendation is to use AI itself to bring people together rather than replace interaction — using the time AI saves on tasks to schedule shared meals, team outings or volunteer activities, and configuring AI tools to route employees back to colleagues rather than always providing direct answers.

"The easiest and most direct way to solve for loneliness is to use AI to create more interaction between people," she says. "That doesn't have to be overly complicated."

The HBR article also recommends that organizations establish clear guidelines for when human interaction should take precedence over AI — keeping coaching, mentoring and conflict resolution as primarily human functions — and that employers invest in "digital wellness" programs to help employees recognize the signs of overreliance on AI for emotional support.

"This is really a leadership opportunity for HR because nobody is minding employee well-being as much as HR,” says Hadley. “The intersection between AI and employee well-being is something HR should own and really lead on — because I don't think it's happening otherwise."