970x125
Chatbots are becoming increasingly sophisticated. And they are getting better and better at creating an emotional bond with the user. This, of course, has its risks, so much so that there has been a lot of talk about regulation.
We easily form emotional bonds with chatbots, and they vary widely from person to person. Some people find it easy to connect with them emotionally, while others find it more difficult. Young people and men tend to belong to the former group, whereas older people and women are more likely to belong to the latter.
A recent study examined these variations in emotional connection to chatbots. Researchers at the University of British Columbia studied the relation between this trait and our tendency to anthropomorphise in 1274 individuals. But what does it mean to anthropomorphise?
Anthropomorphism is our tendency to describe not only AI systems, but any technology in terms of human-like mental states. We are clearly prone to doing so, and this has long been true, well before the advent of chatbots. When you say that the tablet doesn’t want to play the video, that the computer is still thinking, or that your new phone is erratic, this amounts to anthropomorphism. You attribute human-like mental states (like wanting something or being erratic) to a gadget.
The concept of anthropomorphism was not originally introduced to understand our attitude to technology, but rather to describe our default mode of relating to animals. In this context, anthropomorphism is the projection of human emotions, thoughts, and psychological traits onto animals. This is clearly something we are prone to do. However, according to the 20th-century consensus in biology and ethology, it is an attitude we need to suspend when researching the animal mind (although this consensus has been recently broken and there have been attempts to resurrect at least some versions of anthropomorphism as scientifically acceptable).
Anthropomorphism, given its illustrious history, has been studied extensively, and there are various methods for assessing one’s tendency to anthropomorphise. A recent study at the University of British Columbia found a very close correlation between these anthropomorphising tendencies and the degree of emotional and social bond one forms with chatbots.
In some sense, this is not an unexpected result: emotional connection to an entity (a piece of technology or an animal) presupposes that we think of such an entity as something like us. But this entails attributing human-like mental states to it—beliefs, desires, emotions, and wishes.
On the other hand, understanding how our bond with AI systems, in general, is rooted in our anthropomorphising tendencies can help us to monitor and regulate our behaviour towards chatbots. And, as recent incidents show, this can literally save lives.
