Can chat bots really be empathetic?
Last week I was reading some articles in a Linkedin group regarding the growth of chatbots usage in healthcare field, adn that make me think: will patients accept to be treated by Artificial Intelligence systems instead of doctors? And more precisely, by chat bots?
We can appreciate two different but correlated factors to answers these questions: which features do they need to have? How can patients be convinced to use them? The second one depends on the first one. If a chatbot can solve a health problem then people will start using them.
Criticism to healthcare system:
When it comes to illness detection, one of the main criticisms to the actual healthcare systems is that patients must wait a long time to get their first appointment with the doctor, and that the first interaction most likely will redirect the patient to another doctor, the specialist that will be the one able to give a diagnostic.
Chatbots can play here a key role in early illness detection and diagnosis avoiding a waiting time that in some occasions is vital. By asking simple questions to the human, the AI system can start matching the information with all the input it has from medical databases, and see if it can find an illness that fit the given answers. Based on the results, the bot will propose the patient the proper specialist he should visit to avoid waiting times, and book the earliest appointment available with him.
But chatbots can be used as well for following up after the patient has left the hospital and he should continue the treatment. According to some studies when patients participate actively in their recovering processes they heal faster and incur in lower costs. So using bots, may help healtcare systems to save huge amounts of money as well.
The empathy issue:
Healthcare is a concern for all of us. Just the possibility of being sick scares us so we cannot talk about it as if we were talking about the weather or about ordering a pizza. In fact, when we go to the doctor we expect understanding or emapthy from the doctor, and this is exactly what we fear about bots. They are not humans, they are machines so they are not empathetic.
But, if we are able to teach machines all type of words and phrase structures, why can’t we teach them to communicate being empathetic?
The true is that, in fact, we can.
One of the main issues of training is the lack of training materials, usually bots have only received as an input formal structures and formal words, because for so long the paradigm was this was the only information needed. But to communicate as a human, bots need to know how humans communicate with each other, and when it comes to health issues formal language is not the adequate one.
In this traditional training method the only way in which bots can learn is from the experience, by interacting with humans, but… in healthcare no one wants to be part of the bot’s training process, they just want their problems solved.
Thinking about this, obviously we need a new solution, we must launch a bot that can communicate with patients being empathetic from the very beginning and, if it can learn more while interacting with humans it will be even better.
