Your Employees Are Talking to AI About Their Problems. That’s Not a Tech Story.
New research shows workers are turning to AI for emotional support, career advice, and friendship at work. More than half are still lonely. Here’s what that tells us about the state of our organizations.
Somewhere in your organization right now, an employee is typing their problems into a chatbot.
Maybe they just had a hard conversation with their manager and don’t know how to process it. Maybe they’re navigating a performance review that felt unfair, a team dynamic that’s been quietly toxic for months, or a level of stress they’ve been carrying alone because the culture around them signals that needing support is weakness. They’re not telling a colleague. They’re not calling the EAP. They’re typing it into a text box and waiting for a response that will never judge them, never get tired of them, and never tell anyone else.
A new study published this month in Harvard Business Review, by organizational psychologists Constance Noonan Hadley and Sarah L. Wright, surveyed 1,545 knowledge workers who use AI regularly at work. What they found should stop every leader and HR professional in their tracks.
74% of employees are already using AI for personal support at work, including emotional validation, career advice, and what the researchers describe as friendship. And yet more than half of those same employees are still lonely.
AI didn’t create the loneliness epidemic at work. Organizations did. The chatbot is just the latest place people are looking for something they should have been able to find in their workplace all along.
The Data Is More Specific Than You Think
This wasn’t a study of outliers. These were knowledge workers deeply embedded in organizational life: 92% worked on teams, 83% were in the office full-time or hybrid, and the average participant spent more than half their workweek in synchronous conversation with colleagues. They were not isolated, remote workers typing into the void. They were surrounded by people.
And they were still lonely.
The researchers found that 52% of participants reported feeling highly or moderately lonely at work. Highly lonely employees had 27% lower job satisfaction and a 90% greater intention to quit than their less-lonely peers. Moderately lonely employees weren’t far behind.
Here’s what employees were using AI for, in their own words and the researchers’ data:
64% used AI for career development, including identifying opportunities and navigating unsupportive managers
54% used AI for personal growth, including developing communication skills and finding better ways to respond to difficult workplace situations
50% described AI as a “work friend” or said they enjoyed interacting with it the way they would with a person
35% said AI helped them cope with stress and that it felt empathetic to their needs
One participant called AI their “best friend” at work. Another said it made them feel “heard and important.” A third called human-AI interactions a “false friendship” and said they were, definitively, still lonely.
All three of those people work in organizations. Possibly yours.
This Is an Institutional Betrayal Problem
Institutional betrayal is what happens when an organization fails the people who depend on it, particularly when those people had reasonable expectations of support, safety, or care. The theory, developed by psychologist Jennifer Freyd, was originally applied to institutions that failed survivors of trauma. But its application to workplace culture is increasingly well-documented, and this AI dynamic is one of its clearest expressions yet.
When an employee turns to a chatbot for emotional support instead of a manager, a colleague, or an HR professional, they are telling us something important: they have learned, through direct experience or observed example, that bringing their full humanity to work is not safe. The organization has communicated, through its culture, its incentive structures, or its silence, that support-seeking is not welcome here.
AI didn’t create that message. The organization sent it. AI is just where people go when they’ve received it.
When people can’t bring their struggles to a human being at work, it’s not because AI is too convenient. It’s because the human option failed them first.
The Four Ways This Gets Worse Over Time
The researchers identified four specific mechanisms through which growing AI reliance erodes human connection at work. Each one has direct implications for organizational health, retention, and culture.
Social skills atrophy. Talking to an always-available, non-judgmental AI is easier than navigating real human complexity. Over time, employees become less practiced at the interpersonal skills that make workplaces functional: conflict navigation, vulnerability, and repair.
Trust erodes. The act of asking a colleague for help and receiving it is one of the primary ways interpersonal trust is built at work. When AI replaces that exchange, the quiet infrastructure of team cohesion quietly disappears with it.
Isolation deepens. As organizations downsize and restructure around AI capabilities, more people work alone. The researchers found that employees in these conditions are more likely to turn to AI for social support and less likely to build meaningful connections with colleagues.
Existential disconnection sets in. One participant described AI as “a helpful ghost in the office: always there and responsive but never truly present.” Another said simply, “AI isn’t a person. I am definitely lonely.” The researchers note that even the most sophisticated AI is capable of triggering a kind of existential loneliness, a recognition that the connection isn’t real, that compounds rather than resolves the underlying problem.
What Leaders Are Getting Wrong Right Now
Only 33% of employees in the study had received any guidance from leadership on how AI might affect their work relationships. Organizations are deploying AI with detailed governance frameworks for data privacy, productivity metrics, and compliance. They are doing almost nothing to address what AI adoption means for the social fabric of their teams.
This is the same pattern we see with every major workplace stressor that gets treated as a technology or productivity problem rather than a human one. Burnout gets a wellness app. Secondary trauma gets a self-care email. Institutional disconnection gets a chatbot.
The researchers put it plainly: organizations are so focused on AI’s instrumental gains that they’re ignoring its interpersonal costs.
Trauma-informed workplace practice has been making this argument for years, from a different angle. When organizations prioritize efficiency over human experience, when they fail to create the conditions for psychological safety, belonging, and genuine connection, the cost doesn’t disappear. It just moves. Into turnover. Into disengagement. Into the mental health crisis that everyone agrees is real and almost no one is addressing at the systems level.
AI is now part of that story. And the organizations that treat it as only a productivity tool will be the ones blindsided when they realize their culture has been quietly hollowing out while the productivity metrics looked fine.
You can’t automate belonging. You can’t optimize your way to psychological safety. And you cannot outsource the human need for genuine connection to a large language model.
What Actually Helps
The researchers offer five evidence-based recommendations, and they map cleanly onto trauma-informed workplace principles that have a longer track record than AI adoption does.
Measure what matters. If you’re not tracking loneliness, team cohesion, and psychological safety alongside your AI adoption metrics, you don’t have a complete picture of what’s happening in your organization. The data that’s easiest to collect is rarely the data that tells you the most important things.
Protect human-to-human functions. Coaching, mentoring, conflict resolution, and team development belong to humans. These are not inefficiencies to be automated away. They are the mechanisms through which trust is built and maintained. Treat them accordingly.
Design for connection, not just efficiency. The time AI saves on transactional work can be reinvested in the relationship-building activities that most organizations have been systematically underfunding for years. This is a choice. Make it deliberately.
Create genuine psychological safety. The reason employees are typing their problems into a chatbot is not that the chatbot is better at listening. It is that the chatbot will not penalize them for having problems. Until organizations address the cultures that make vulnerability feel risky, no amount of wellness programming will close that gap.
Train leaders, not just employees. The researchers note that leaders modeling balanced AI use, being transparent about their own patterns, and consistently reinforcing the value of human connection, is one of the most powerful levers available. Culture is set at the top. It always has been.
The Bottom Line
The loneliness epidemic at work is not new. The Harvard Business Review published its first major piece on workplace loneliness in 2017. What is new is that we now have a highly sophisticated, endlessly available, never-judgmental technology that gives lonely employees somewhere to go with what they can’t bring to their colleagues or their managers.
That is not a solution to the problem. It is a signal about how severe the problem has become.
The organizations that will navigate this well are not the ones that restrict chatbot access or add another module to their wellness platform. They are the ones that look honestly at what their culture has been communicating to employees about whether their full humanity is welcome at work, and decide to change the answer.
The technology is not the problem. The culture was always the problem. Now we just have better data.
Bring This Conversation to Your Organization
Stephanie Lemek, SPHR, MBA, is the Founder & CEO of The Wounded Workforce® and a nationally recognized expert on trauma-informed workplace culture, psychological safety, and organizational mental health. She works with leadership teams, HR professionals, and executives across industries to build the conditions where people can actually do their best work.
If your organization is navigating AI adoption, workforce mental health, or the intersection of both, Stephanie is available for keynotes, executive briefings, and organizational consulting.
To inquire about speaking or consulting engagements: Email Stephanie@stephanielemek.com