How Lenny the bot convinces you that he is a person: Storytelling, affiliations, and alignments in multi-unit turns

Abstract

This research delves into the world of conversation analysis, focusing on the unique conversational agent, Lenny. In contrast to most modern AI-based chatbots, Lenny employs a minimalistic approach, utilizing pre-recorded turns to engage with unsolicited callers and extend interactions. The study aims to dissect how Lenny’s long turns contribute to displaying ‘his’ personhood. By analyzing Lenny’s long turn in interaction, we uncover how it consolidates Lenny’s relatable character. Through analysis of a corpus of recorded interactions, the paper highlights the role of turn design in simulating human-like interactions. Ultimately, this research offers insights into the interplay between scripted content and human understanding in live conversations.