Anthropomorphism of chat interfaces: Why you can’t get away with not designing a chat interface personality
A study was completed for my masters thesis (IADT MSc UXD) to determine if the personality of a chatbot had an effect on user satisfaction and engagement. Two chatbots were created, Chatbot A and Chatbot B. Chatbot A was casual and enthusiastic, whereas Chatbot B was formal and very serious (some might say that it had no personality).
The quantitative results determined that there was a marginal difference between the interaction with Chatbot A and Chatbot B. The time spent with both on average differed by 7 seconds. The measurement of satisfaction was the exact same.
However, when interviewing participants after the tasks had been completed, users reflected that they felt as if Chatbot A felt like a ‘friend’ and Chatbot B was a ‘robot butler’. Some users even came back to chat to Chatbot A up to a month after the tests had taken place.
Even if the chat interface creator tries purposely to not have a name or a personality type associated with a chat interface (as I did for the study), it is still human nature to anthropomorphise things in order to make sense of them.
The talk will also go into some detail about biases of humans towards perceived gender of chat interfaces and how we should be designing to challenge them.
I’m in the process of completing more work with chat interfaces at the moment and gathering information on studies to see how humans interact with bots.