Leveraging Artificial Intelligence (AI) As a Critical Friend: The Affordances and Limitations

Abstract

This qualitative self-study contributes new insights on critical friendship, specifically related to Artificial Intelligence (AI) chatbot utility trends, asking: In what ways can AI serve as our critical friend? And, what are the affordances and limitations of AI as a critical friend, compared to our human critical friends? To answer these questions, the authors selected Pi.ai/talk, an AI tool they later nicknamed Pia. After interacting with Pia (using data from a previous study), they compared Pia's responses to those of two human critical friends, Anne and Anders, who originally provided feedback in their previous self-study. Drawing on theoretical and conceptual perspectives (Thought Collective, Anatomy of Critical Friend Attributes, & Critical Friend Definition Continuum) and using four data sources, the authors applied multiple qualitative analysis tools to determine the affordances and limitations of Pia. They found Pia could serve as a critical friend when examining her critical friend attributes because to some extent she provided feedback to their thinking, probed thoughtful questions, and remained in a positive tone. Although Pia had some affordances in her ability to serve as a critical friend, the data evidenced several limitations in how she met the critical attributes in the areas of constraints, trust, and relational dynamics; especially when juxtaposing Pia's feedback to Anne's and Anders' feedback. Given the findings, the authors argue AI tools can complement human critical friends in self-study research, but as of the date of this publication, they do not advocate for replacing human critical friends entirely.

Ähnliche Arbeiten