When we design an AI assistant, we also shape how it comes across. A concise answer, an enthusiastic greeting, or an accommodating tone can change what people expect from the system.
Vibe Check studied that relationship in a travel-planning task with 150 participants. Agents with moderate expression across the Big Five personality traits received the most favorable evaluations overall. Greater user–agent alignment was also associated with more positive perceptions.
The lesson is contextual: the study offers evidence for a particular task and setting, not a universal personality for every agent. It opens questions about how designers can make personality choices explicit and responsive to people’s needs.