One of the most challenging obstacles facing chatbot developers is getting chatbots to seem more personable and empathetic. When most people think of talking to a computer program, they imagine a cold and unfeeling AI. If brands what people to interact with their customer service chatbots, they’re going to need to be able to sympathize with customers, pick up on emotional cues and respond in an appropriate manner. Human customer service agents do this without thinking, for chatbots it’s much harder. A team of researchers at Northeastern University in Boston is working on a chatbot that is programmed to deal with one of the most emotionally sensitive topics of all: death. While their goal isn’t a more emotionally intelligent customer service chatbot, their work may have some implications for customer service bots in the future.
Filling a need
Professor Timothy Bickmore who heads the project said the idea was first conceived when he noticed a need. Many older, terminally ill patients are lonely and scared in their final days, weeks, and months. They don’t always have a loved one to guide them through many of the difficult decisions and preparations that need to be handled prior to death and so the concept of an end of life chatbot was conceived.
What it can do
The end of life chatbot can assist terminally ill patients in a number of different ways. It is programmed to walk people through information regarding power of attorney and creating a living will so that they understand what options they have should they become incapacitated and kept alive by medical apparatuses.
The chatbot can guide patients through meditation sessions to help keep them calm if they are feeling afraid. It can also discuss death and the beyond in philosophical or religious terms depending on the patient’s preference. It can even chat with patients about a wide range of topics with the purpose of keeping them company.
In order to create such a chatbot, Bickmore’s team consulted doctors, hospital chaplains, and AI researchers to develop a better understanding of what an end-of-life chatbot would need to be able to do for the people it converses with.
Further testing is still needed but in early trials, the majority of those who communicated with the end-of-life chatbot reported feeling more at ease and more prepared as they faced death.
Bickmore’s work suggests that chatbots can, to some degree, connect with users on an emotional level and even bring about emotional responses in those they chat with. It’s an important development as more businesses try to create chatbots that can better simulate an interaction with a human customer service agent.