Social chatbots and character: Mind-modelling, mimetics, and postdigital misperception

Abstract

This article examines how users cognitively process social chatbot output as a novel form of participatory, generative “character,” co-produced through human–AI interaction. Reporting on the results of empirical research on user responses to the social chatbot platform character.ai, it shows how cognitive stylistic and narratological approaches to character can be utilised but also must be developed to analyse and theorise users’ experience of and interaction with generative AI agents. It makes specific contributions to cognitive stylistics, rhetorical narratology, character studies, and research on gen AI chatbots by: (1) illustrating how the concept of mind-modelling can used to analyse the ways in which users linguistically anthropomorphise social chatbots; (2) developing two new audience response types – transmimetic and hypermimetic – to respectively capture ways in which users assess chatbots against prior knowledge of characters across media and cases where chatbots appear so convincingly human that users question their ontological status; (3) proposing the concept of ‘postdigital misperception’ to reflect the way that users of AI can misattribute the ontological origin of an LLM’s output. The article demonstrates how social chatbots are reshaping how users understand character, agency, and reality in interactive, participatory and generative digital contexts and argues that social chatbots epitomise postdigital experience in blurring the distinction between the digital and non-digital as well as fictional and real.