Among all great speakers, panels and keynotes that we assisted during the ChatBot Summit in Berlin, one of the most spectacular ones, in my opinion, was the keynote delivered by Prof. Justine Cassell “Winning Friends and Influencing People with Bots”.
With her keynote speech about chatbots and the future of human-bot interaction Prof. Cassell managed to shed some light on what is really the status-quo of current human-chatbot interaction, and shared the significant amount of research that is taking place at the Carnegie Mellon University, as well as showing the audience a brief interaction with SARA (Socially Aware Robot Assistant), in order to take this interaction at the next level.
One important thing that Prof. Cassell mentioned, and we as chatbot developers have to agree with, is the fact that right now, unfortunately, what we think of a true conversation with a chatbot, is actually more like the interaction between us and a search engine. However, it is in a different setting. What all current technology is missing is the “social awareness factor”. To put it in a simple way, you talk completely differently with a friend that you know for 10 years than what you do with a stranger you just met at a party. You need to have appropriate responses to appropriate situations.
The amount of research done at the CMU with regards to this topic is staggering! Looking at human to human interactions, collecting natural data, finding proper ways of interpreting that data. For bots to be as humane as possible, all the interpersonal functions and relationships should be considered.
Last but not least, what the current chatbots on the market are lacking, are the Temporal Association Rules, a feature crucial in any human to human conversation.
Now, why is the research that is going on at the CMU so valuable for the chatbot industry? Studies show that when people interact with bots that are more relatable, they have increased task performance and spend more time chatting with them.
Verbal and texting behavior in communication might seem different at the first glance, however, studies show that people usually use the same features in texts as in verbal behavior, so with more research identifying tones and emotions that are emitted by a person via chat will be possible too.
As an end note, we do wish CMU and Prof. Cassell best of luck in their ongoing research of human-machine interaction, and we hope that their findings will help advance the chatbot industry to the next level.