Electronic Brainstorming With a Chatbot Partner: A Good Idea Due to Increased Productivity and Idea Diversity

Abstract

Brainstorming is a creative technique that fosters collaboration to enhance idea generation. The occurrence of evaluation apprehension, a fear of being evaluated negatively by others, however, can stymy brainstorming. How the advantages of collaboration can be leveraged while evaluation apprehension is prevented is an open scientific and practical problem. In this brief research report, it is proposed that chatbots could provide a solution. Chatbots can be designed to share ideas with their users, facilitating inspiration. Compared to human beings, chatbots are also perceived as possessing limited agency and evaluative capacity. This could reduce evaluation apprehension. Given that chatbots are often embedded in a textual chat interface, social cues (picture, name, and description) can reinforce the perceived chatbot identity, enhancing its alleged effects on evaluation apprehension and subsequently on brainstorming performance. These conjectures were tested in an online 2 × 2 between-subjects experiment (n = 120) where people were instructed to brainstorm with a partner that was framed as either a chatbot or human being (but followed the same automated script), with or without the presence of social cues. The results showed that brainstorming with a chatbot led participants to produce more ideas, with more diversity than brainstorming with an alleged human being. Social cues enhanced the effect on idea diversity, but only with the chatbot. No significant effects on evaluation apprehension were found. The contribution of this study is therefore that chatbots can be used for effective human–machine teaming during brainstorming, but this enhancement is not explained by its effects on evaluation apprehension.

Cite this work

No public copy is stored for this record, so only the metadata is cited.

Stored only when both papers are in the corpus — 44 citations exist upstream.

Papers that cite the same references — the relational fallback while embeddings are still backfilling.