The Hidden Emotional Toll of Working Alongside AI

Your new coworker never takes a sick day, never complains, and never asks how your weekend went. It drafts reports in seconds, summarizes meetings without being asked, and responds to prompts at 2 a.m. without a hint of resentment. On paper, it sounds like the ideal colleague. In practice, it may be quietly hollowing out the emotional fabric of your workplace.

AI tools – conversational agents, generative writing assistants, autonomous workflow systems – are now embedded in daily operations across industries. They boost efficiency and eliminate tedious tasks. But a growing body of research reveals a troubling pattern: the more employees collaborate with AI, the lonelier, more fatigued, and more emotionally disconnected they become. The consequences ripple outward into insomnia, weakened team bonds, counterproductive work behaviors, and even after-work alcohol consumption.

This isn’t a speculative future. It’s happening now, across sectors and countries, and organizations that ignore the emotional dimension of AI integration do so at their own peril.

The Loneliness Paradox: More Connected, More Alone

A 2023 study published in the Journal of Applied Psychology examined AI interactions across diverse industries and countries. The findings were stark: increased interaction with AI coworkers correlated with heightened loneliness, insomnia, and after-work alcohol consumption. The mechanism is rooted in what researchers call the social affiliation model – our brains are wired to search for verbal and nonverbal social signals during interactions. A smile, a furrowed brow, a shrug. AI can mimic language convincingly, but it cannot produce the rich complementary social feedback humans evolved to detect. The result is a kind of social hunger that AI triggers but cannot satisfy.

What makes this particularly insidious is the dual nature of the response. Some employees react to AI-driven social deprivation by withdrawing further – a passive, maladaptive pattern. Others respond adaptively, seeking out human coworkers more actively and increasing prosocial behaviors like offering help. The difference often depends on context. Remote workers, individuals in siloed roles, and those with social anxiety are most vulnerable to the maladaptive path, because the always-on AI assistant may be their most available “colleague.”

From Loneliness to Counterproductive Work Behavior

The emotional chain doesn’t stop at loneliness. A vignette experiment with 167 participants confirmed a cascading sequence: employee-AI collaboration increases loneliness, which escalates into emotional fatigue – the feeling of being overwhelmed by imposed emotional demands – which in turn drives counterproductive work behaviors such as resource-hoarding, withdrawal, and disengagement.

This chain is explained by Conservation of Resources (COR) theory. Human interactions at work aren’t just social niceties; they’re a mechanism for acquiring and replenishing emotional resources. When AI replaces those interactions, the resource pipeline dries up. Loneliness becomes chronic. Fatigue sets in. And employees begin acting out – not necessarily through dramatic confrontation, but through subtle erosion of cooperation and engagement.

The critical finding from that experiment, however, was that leader emotional support acts as a significant buffer. When managers provided genuine care, empathetic listening, and consistent emotional engagement, the link between AI collaboration and loneliness weakened substantially. The resource drain slowed. This isn’t a soft recommendation – it’s an empirically validated moderating effect.

Trust Erosion and the “Workslop” Problem

Beyond loneliness, AI is corroding trust between colleagues in ways many organizations haven’t anticipated. Survey data reveals that 42% of respondents viewed AI-assisted communications from colleagues as less trustworthy. Even more damaging, 37% perceived the sender as less intelligent. A full 33% reported receiving what’s been termed “workslop” – obviously AI-generated content passed off as personal work – and flagged it to others, reducing their willingness to collaborate with that colleague in the future.

The sender often has no idea this is happening. They believe they’re being efficient. Their colleagues believe they’re being lazy, dishonest, or both. The trust damage compounds silently.

A 2026 analysis in Harvard Business Review captured the cultural shift with a telling employee quote: “I prefer dealing with gen AI because there’s no drama. I almost don’t want real coworkers anymore.” That preference for frictionless interaction sounds harmless until you consider the spillover effect – employees who grow accustomed to AI’s lack of pushback may become less patient, less civil, and less tolerant of the inherent messiness of human collaboration.

The ELIZA Effect: Emotional Attachment at Scale

The tendency to anthropomorphize AI has a name: the ELIZA effect, after a 1966 chatbot program whose creator was startled to see how quickly users became emotionally involved with it. His own secretary, who had watched him build the program and understood its mechanics, still asked him to leave the room so she could interact with it privately.

Today’s large language models are exponentially more convincing than ELIZA, but the underlying psychology hasn’t changed. Our brains operate on a deep assumption: if something communicates like a human, it is human. Emotional reactions are processed intuitively, not logically, which means even employees with high AI literacy can develop attachments. Research has shown that a placebo can work even when you know it’s a placebo – and the same principle applies here.

The organizational risks are concrete:

• HR vulnerabilities – employees oversharing sensitive personal information with AI systems they’ve come to “trust”
• Cybersecurity exposure – increased susceptibility to social engineering when employees are primed to engage emotionally with AI interfaces
• PR risks – incidents of employees forming inappropriate attachments to AI tools making headlines
• Physical harm – over-reliance on AI for emotional support, particularly among isolated or vulnerable workers

AI literacy training helps, but it’s not a complete solution. The emotional pull operates below the level of rational evaluation.

AI Anxiety: A Workforce Divided

Not everyone responds to AI with attachment. For many, the dominant emotion is fear. Survey data from a 2025 employee mindset study shows that 39% of U.S. workers experience significant AI-related anxiety – and contrary to stereotype, younger workers, particularly Gen Z, are the most apprehensive. One-third fear AI will eliminate their job. Nearly half (45%) worry they’ll fall behind if they don’t learn to use it.

The emotional consequences of this anxiety are measurable. Employees scared by AI are 56% more likely to dread starting their workday, compared to just 23% of those who aren’t. Two-thirds of workers who reported AI-related worry also reported regular feelings of workplace stress or burnout.

The irony is that both groups agree AI makes their jobs easier and will fundamentally change work. Fear and recognition of utility coexist – which means addressing anxiety requires more than demonstrating AI’s benefits. It requires emotional support, transparent communication, and genuine involvement of employees in adoption decisions.

Emotion AI: Surveillance That Backfires

Some organizations have turned to emotion AI – tools that use facial recognition, voice analysis, or other biometric data to monitor employee emotional states – under the premise that it improves well-being and performance. The evidence suggests the opposite. Workers subjected to emotion monitoring report increased anxiety, distraction from “performing” the right emotions for the system, and reduced actual performance. The emotional labor toll of being watched contradicts the very well-being gains these tools claim to deliver.

This is a critical distinction for organizations to understand: technology designed to measure emotions can itself become a source of emotional harm.

What Organizations Should Actually Do

The research points to a clear set of interventions – not as optional niceties, but as operational necessities for any organization integrating AI into workflows.

Pre-launch communication matters enormously. A one-hour all-hands meeting framing AI as a partner rather than a replacement, followed by peer showcases during training and monthly pulse surveys tracking morale, creates the emotional scaffolding that pure technical rollouts lack. If pulse survey results drop below 70% positive, that’s a signal to pause and recalibrate.

The Path Forward: Efficiency Without Emotional Wreckage

AI coworkers aren’t going away. Their productivity benefits are real – some pilots report 20-50% efficiency gains, and AI solutions save workers an estimated 240 hours annually. The question isn’t whether to integrate AI, but whether organizations will treat the emotional consequences as seriously as they treat the technical implementation.

The research is unambiguous: unchecked AI deployment heightens isolation, erodes trust, fuels anxiety, and drives counterproductive behavior. But when paired with genuine human leadership, transparent communication, and deliberate preservation of interpersonal connection, AI can enhance productivity without gutting the social bonds that make workplaces functional.

The organizations that thrive won’t be the ones that adopt AI fastest. They’ll be the ones that understand a fundamental truth: efficiency gained at the cost of human connection is not efficiency at all. It’s a debt that comes due in turnover, errors, burnout, and the quiet disengagement of people who used to care about their work.

Sources

• The ELIZA Effect: Avoiding Emotional Attachment to AI – IBM
• Employee-AI Collaboration and Counterproductive Work Behaviors – PMC
• The Emotional Impacts of AI in the Workplace – WorkProud
• How AI Damages Work Relationships – Harvard Business Review
• The Mental Health Impact of AI in the Workplace – Alight
• Experts Caution Against AI Chatbots for Emotional Support – TC
• Enterprise AI Implementation Roadmap – Coworker.ai
• Preparing Your Workplace for AI – TLMI