Conversational agents are often used in public organizations as an entry-level application of artificial intelligence and constitute a form of digitally mediated public encounter. However, little is known about how citizens communicate in conversational-agent-mediated public service interactions. By analyzing conversations between citizens and a conversational agent, this article formulates three propositions that contribute to understanding the communicative dynamics of such encounters.
Data in the form of conversations between citizens and a GPT-based conversational agent collected during an experiment are analyzed in a qualitative manner (N = 73). In the experiment, study participants had to fill in a fictitious tax return and were thereby assisted by a conversational agent.
Results show that citizens engage in highly task-oriented communication, seeking answers through simple and concise questions rather than elaborate prompt engineering. The conversational agent is expected to have a lot of knowledge in the topic in question. Even if tried to be avoided, in rare cases, the conversational agent provided incorrect but plausible answers (hallucination). These findings suggest that conversational-agent-mediated public encounters are characterized by strong expectations regarding information provision, limited user engagement with prompting strategies and vulnerabilities arising from the agent's role as an information intermediary.
This study is among the first to analyze conversational log data from citizen interactions with a GPT-based conversational agent in a public-service context. By examining these interactions as conversational-agent-mediated public encounters, it contributes empirical evidence to emerging research on digital public encounters and citizen–state communication.
