"Looks Like Robots, Sounds Like Humans": Surveying Students' Conceptualizations of Learning Agents (Poster 36)

• In Event: AERA Virtual Poster Session 2
  In Poster Session: Designing Meaningful Learning Experiences

Abstract

Conversational agents, systems that employ natural language understanding to give learning support, can be integrated into educational contexts. Understanding students’ conceptualizations of conversational agents is important to reduce misalignments of expectations and engage learners. We asked 52 high school students (ages 14-15) to describe in writing their ideal learning agents. We found that students varied in their descriptions of the agent’s forms, such as a human, robot, animal, or inanimate objects. Most students referenced smart-home conversational agents and desired human characteristics such as personalities, multimodal output, broad-domain intelligence, and social interactions. Findings illustrate how students draw from their everyday experiences to conceptualize conversational agents and inform future designs to integrate human-like features into agents to improve learning interactions.

Author

• Ha Nguyen, Utah State University