The Problem with Talking Refrigerators
According to a new report published by scientists Adam Waytz from Harvard University and Nicholas Epley and John T. Cacioppo from University of Chicago, humans have no difficulty identifying other humans in a biological sense, but from a psychological standpoint, things are a lot more complicated. Their report analyzes the psychology of anthropomorphism, which is basically our tendency to attribute human characteristics or behavior to non-human things. We express this tendency in a variety of ways, ranging from dressing up pets as people, attaching names to natural phenomenon, inanimate objects such as Roombas or for that matter, human-like gods.
According to the research carried out by Waytz and his colleagues, thinking of a non-human entity in human ways helps us not only feel more in control and connected to our environment, but also renders non-human objects and phenomenon worthy of moral care and consideration.
In other recent news, McKinsey recently released a report about “new sensor-driven business models” that the Internet of Things will bring. According to the report, we will soon see the emergence of Internet-enabled objects ranging from shoes to refrigerators that layer web services, which will help us interact with these objects. This will be made possible by technologies such as RFID tags and sensors, resulting in numerous new technological and social advances.
These two recent reports made me realize that it’s possible we’re at the precipice of a new wave of anthropomorphic computer interface design. We already spend so much time interacting with computing devices that are connected to the Internet, but imagine if objects all around us were also networked and interactive. In such a world, anthropomorphic human-computer interaction design may actually make things simpler and more intuitive for us. The use of anthropomorphic agents in human-computer interaction is fairly common, but has yielded varied results. On the plus side, they have the potential to trigger an emotional response from users, improve interactivity, engagement and thereby promote broad based adoption of computer technology. However, human-like agents can also be incredibly annoying (I’m thinking of Clippy, the Microsoft Office Assistant). 
Unfortunately, annoyance is not the biggest potential problem with human-like computer agents. As the research conducted by Waytz and co. has pointed out; imparting human-like qualities on inanimate objects illicit subconscious and psychological responses in humans. Although their research touches upon some of the positive psychological drivers for anthropomorphism, it does not delve into the responses that are evoked by actual human-like interfaces such as robots that simulate human appearance, expressions and gestures, or human-like interfaces in cars, websites, software programs and mobile devices. If anthropomorphism helps us interact and control inanimate objects with ease, what negative impacts can such human-like interfaces or objects have on us? This paper provides some very interesting insight into some of the privacy implications of anthropomorphic human-computer interaction design techniques. However, privacy will not be the only concern and such interfaces could have an impact on a range of human values from individuality, respect, empathy and trust to informed consent, authority and security. I really don’t mean to be a buzz-kill, but I think the “Internet of Things” could use some principles of value sensitive design. If we don’t identify, understand and anticipate the ethical and value concerns that are bound to arise from the deployment of networked objects that exhibit human-like behavior, we risk facing an avalanche of problems later.
So anyone out there working on building a smart, human-like refrigerator – I hope you’re paying attention.
