Your New Manager Is AI: How AI Leadership Style and Physical Embodiment Shape Human Performance and Well-Being in the Workplace

Artificial intelligence is rapidly entering our workplaces. Intelligent chatbots answer our questions, help draft emails, and assist with complex tasks. Soon, AI-powered humanoid robots may work alongside us in offices. But as AI systems take on roles previously held by human supervisors and collaborators, a crucial question emerges: How should AI interact with us to help us work effectively while maintaining our well-being?

We know from research that a human boss's management style profoundly affects employees. Supportive managers who encourage autonomy and express confidence in their workers tend to foster creativity and job satisfaction. Controlling managers who micromanage and issue strict commands may achieve quick results but often at the cost of employee motivation and well-being. Do these same principles apply when our "manager" is an AI? And does it matter whether the AI has a physical body - a humanoid robot standing beside us - or exists only as a chatbot on our screen?

This project investigates how the interaction style of AI workplace assistants affects human performance, emotions, and stress levels. We will study 240 office workers in Poland and the United States as they complete realistic work tasks - writing memos, composing emails, and brainstorming ideas - while supervised by an AI assistant. Half will interact with a supportive AI that encourages their input and respects their autonomy; half will work with a directive AI that gives strict instructions and closely monitors their work. We will also compare interactions with a physical humanoid robot versus a chatbot on a screen to determine whether physical presence amplifies these effects.

The cross-cultural comparison between Poland and the United States is particularly important. These countries differ significantly in cultural attitudes toward authority and hierarchy - Polish culture traditionally accepts more hierarchical relationships, while American culture tends to emphasize equality and individual autonomy. By studying participants from both countries, we can discover whether effective AI design needs to be tailored to different cultural contexts or whether universal principles apply.

To capture the full picture of human responses, we will measure physiological stress indicators (heart rate, skin conductance, facial muscle activity), analyze conversation patterns, have experts evaluate work quality, and collect participants' own assessments of their experience. This comprehensive approach allows us to detect not only what people consciously report but also how their bodies respond to different AI interaction styles.

Our preliminary studies have already shown that AI interaction style dramatically affects human behavior - participants working with supportive AI showed more confidence and engaged in collaborative negotiation, while those facing directive AI exhibited more frustration, resistance, and attempts to circumvent the system. The current project extends this research to physical robots and cross-cultural settings.

The findings will provide practical guidelines for companies deploying AI assistants, for AI developers designing interaction patterns, and for policymakers considering workplace AI regulations. Our goal is to ensure that as AI becomes our colleague - and perhaps our supervisor - it enhances not only productivity but also human flourishing in the workplace of the future.