Organizations increasingly introduce AI systems that no longer appear as a single assistant, but as a team of specialized agents. For example, one agent may generate ideas, another may critique them, another may check facts, and another may coordinate the workflow. While such systems promise better outputs, they also change how workers interpret AI assistance. Users may perceive agent teams as more competent, more deliberative, more confusing, or more accountable than single AI assistants.

This thesis investigates how knowledge workers make sense of multi-agent AI systems. It asks how users interpret role labels, agent specialization, disagreement between agents, and the apparent coordination of an AI “team.” Based on qualitative interviews, the thesis develops a framework for understanding how workers perceive, trust, and rely on AI agent collectives.

The thesis is supposed to answer the research question: "How do knowledge workers interpret and evaluate multi-agent AI systems that present themselves as teams of specialized agents?"

If you are interested in this master's thesis, feel free to contact Marc Grau (marcchristopher.grau@unisg.ch)