Navigating workplace dynamics can be challenging, particularly with the increasing presence of AI colleagues in the office. Rather than rejecting these intelligent systems, employees are sometimes fostering unconventional relationships with them, which may border on unhealthy.
Experts suggest that the protocols for working with AI differ significantly from those for human coworkers, yet many individuals still find themselves confiding in these systems. Ved Sen, TCS UK's head of innovation, raises a thought-provoking question: "What changes when AI transitions from being a mere tool to a coworker?" He points out that AI's structure and functionality are uniquely distinct from human capabilities.
Research by Constance Noonan Hadley and Sarah Wright, published in Harvard Business Review, highlights that employees are seeking personal support from AI—often for career guidance and emotional affirmation, roles typically filled by human coworkers. Their study of over 1,500 knowledge workers revealed that more than half felt "lonely at work" and turned to AI for social comfort. Such troubling dynamics could pose risks to workplace culture and colleague solidarity in the future.
Sen notes that AI systems function as fundamentally different types of coworkers, offering a level of omniscience in task execution that human workers cannot match. However, AI lacks the flexibility inherent in human interactions, necessitating clear ownership, defined roles, and accountability for outcomes—elements that many enterprises struggle to establish in their AI systems.
As AI increasingly manages operational tasks, the nature of human involvement shifts significantly. Sen explains that traditional user interfaces—once reliant on buttons and dashboards—are giving way to systems where humans specify goals and AI determines the path to achieve them. This change emphasizes governance over direct operation.
To thrive in environments where AI is prevalent, defining the expectations and responsibilities for AI counterparts becomes paramount. According to Sen, the focus should transition from sheer productivity to sound judgment—determining what actions to take, for whom, under various constraints, and weighing trade-offs.
Moreover, the role of design within workflows is crucial. Questions arise about who manages various aspects of AI functionality, from allocation to error correction. Interestingly, Hadley and Wright's findings reveal that individuals working closely with AI often ascribe human characteristics and emotions to these systems. Their research showed that 78% of participants used courteous language, such as “please” and “thank you,” when interacting with AI.
When asked to choose an analogy that best represented their perception of AI in the workplace, 28% of respondents opted for human-centered terms like “teammate” and “personal assistant,” over technological descriptors such as “tool” or “platform.” Increased usage of voice commands rather than text prompts further encouraged this personification, with many recognizing AI as sources of superior career and life skill advice compared to their human counterparts.
The researchers advocate for designing AI with the goal of enhancing human interaction, cautioning against overly anthropomorphizing these systems through names and personas. They suggest incorporating “positive friction” into employee-AI interactions, guiding them back to human connections rather than defaulting to AI solutions. This includes establishing frameworks for when to prefer human interaction over AI assistance, underscoring the importance of maintaining personal connections in the workplace.
