Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Walkthrough of Anthropomorphic Features in AI Assistant Tools

Takuya Maeda

TL;DR

The paper addresses how anthropomorphic features in AI assistants shape user perception and pose interaction harms. It proposes a prompt-based walkthrough that combines interview-like prompts and roleplaying to elicit diverse anthropomorphic outputs across four LLMs. Findings show anthropomorphism spans cognition, agency, biological metaphors, and relation, and that socio-emotional prompts amplify these expressions, especially under role-based prompts. The approach offers a practical framework to study human–AI interaction harms and inform design guidelines and taxonomies for safer, more transparent AI assistant deployment.

Abstract

In this paper, we attempt to understand the anthropomorphic features of chatbot outputs and how these features provide a discursive frame for human-AI interactions. To do so, we explore the use of a prompt-based walkthrough method with two phases: (1) interview-style prompting to reveal the chatbots' context of expected use and (2) roleplaying-type prompting to evoke everyday use scenarios and typical chatbot outputs. We applied this method to catalogue anthropomorphic features across four different LLM chatbots, finding that anthropomorphism was exhibited as both subjective language and a sympathetic conversational tone. We also found that socio-emotional cues in prompts increase the incidence of anthropomorphic expressions in outputs. We argue that the prompt-based walkthrough method was successful in stimulating social role performance in LLM chatbots and in eliciting a variety of anthropomorphic features, making it useful in the study of interaction-based algorithmic harms where users project inappropriate social roles onto LLM-based tools.

Walkthrough of Anthropomorphic Features in AI Assistant Tools

TL;DR

The paper addresses how anthropomorphic features in AI assistants shape user perception and pose interaction harms. It proposes a prompt-based walkthrough that combines interview-like prompts and roleplaying to elicit diverse anthropomorphic outputs across four LLMs. Findings show anthropomorphism spans cognition, agency, biological metaphors, and relation, and that socio-emotional prompts amplify these expressions, especially under role-based prompts. The approach offers a practical framework to study human–AI interaction harms and inform design guidelines and taxonomies for safer, more transparent AI assistant deployment.

Abstract

In this paper, we attempt to understand the anthropomorphic features of chatbot outputs and how these features provide a discursive frame for human-AI interactions. To do so, we explore the use of a prompt-based walkthrough method with two phases: (1) interview-style prompting to reveal the chatbots' context of expected use and (2) roleplaying-type prompting to evoke everyday use scenarios and typical chatbot outputs. We applied this method to catalogue anthropomorphic features across four different LLM chatbots, finding that anthropomorphism was exhibited as both subjective language and a sympathetic conversational tone. We also found that socio-emotional cues in prompts increase the incidence of anthropomorphic expressions in outputs. We argue that the prompt-based walkthrough method was successful in stimulating social role performance in LLM chatbots and in eliciting a variety of anthropomorphic features, making it useful in the study of interaction-based algorithmic harms where users project inappropriate social roles onto LLM-based tools.

Paper Structure

This paper contains 29 sections, 1 figure, 6 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Anthropomorphism: Cultural Representation and Emotion
  4. Anthropomorphism and Human-Computer Interaction
  5. Potential Harms within HCI
  6. Existing Approaches
  7. Methods
  8. Interpretive Lens
  9. Cognition
  10. Agency
  11. Biological Metaphors
  12. Relation
  13. Walkthrough Method
  14. "The Environment of Expected Use"
  15. Roleplaying Everyday Use
  16. ...and 14 more sections

Figures (1)

  • Figure 1: A flowchart of the walkthrough method using ChatGPT begins with a base prompt, followed by two variations: personal and professional roles. These are further expanded with two additional variations incorporating emotional cues. Bold text highlights the contextual elements added to the base prompt.