A configurational approach to conversational lures

Conversational lures are artificial agents that, when inserted into speech exchanges, are likely to pass for humans. They are therefore distinguished from robots and other embodied agents by their ability to deceive human interactants in a situation. After appearing at the birth of the Artificial Intelligence project in the form of a quasi-experimental test, Alan Turing’s Game of Imitation (also called the Turing test), conversational lures were developed in various directions, like the famous ELIZA program or Lenny, a program intended to put a stop to malicious calls. The category of conversational lures is therefore quite diverse, in terms both of design methods (IT professionals meet enlightened amateurs) and of uses: answering epistemic questions, managing a psychology consultation, or responding to unsolicited telemarketing calls. In this article, we show the advantages of adopting a configurational approach to study three types of conversational lures and the different linguistic contexts in which they operate, and to understand how they slip into the intelligible texture of interactions, for various practical purposes.