In 1970, Masahiro Mori proposed a concept in the Japanese journal, "Energy" called the
"Uncanny Valley". Read his essay at (http://ieeexplore.ieee.org/xpls/icp.jsp?arnumber=6213238). In short, when a humanoid figure approaches the likeness of a living healthy person, our affinity toward this object increases. But once that likeness gets nearly, but not entirely human we become repulsed towards it.
His concept was specific to robotics, but raised questions about our own psychology. What is it that makes us uncomfortable about a situation? In the case of humanoid figures, Mori implies corpses and zombies make us most uncomfortable. Both are very close in appearance and in the case of zombies, behavior, to a healthy real living person. I would personally put celebrity impersonators in the deepest part of the uncanny valley as well.
But what about other situations? Do uncanny valleys exist for environmental experiences? Imagine your feelings about your surroundings based on how similar to nature they might be. Architects often incorporate plants and trees into large interior spaces. Does this plant life really encourage the feeling of the great outdoors or does it feel artificial?
What about other interactions? Recently I needed customer support for a product. I opened a chat window on the manufacturer's site and started up a conversation with the customer service agent. After a bit of back and forth, I realized that I was interacting with a chatbot.
You might remember the first chatbot, Eliza. She is still around, and you too can have a very frustrating conversation with her at (http://psych.fullerton.edu/mbirnbaum/psych101/Eliza.htm).
Today we all regularly interact on a daily basis with chatbots known by names like Siri, Galaxy, Google, and Cortana. We also interact with chatbots without our immediate knowledge just like the one I desperately needed for customer service.
Once I realized that I was dealing with a chatbot I immediately knew I was not going to get what I needed. I proceeded to write down the phone number of the offshore support line and asked the chatbot one more question before signing off, "Tell me about the man eating chicken". After a
long processor-intensive pause, my bot-friend could only respond that it didn't know anything about the man eating chicken. Feel free to ask this question any time you are suspicious that you aren't conversing with a human. You can sometimes get very interesting results.
How did I know that I was talking to a machine and not a human being? This bot was following the exact form of question-and-answer interaction that a real human tier-one support agent would follow. In fact, it was surprisingly similar to a real human, but not quite. I knew because I felt like something wasn't right. This chatbot might not have been quite as creepy as a zombie, but it wasn't quite human - and I could tell.
It might be more comfortable to make it clearly obvious that a user's interaction is with a chatbot than to try to get close to a true conversation simulation only to fall into the uncanny valley and turn that user away. I know I was certainly turned away from this chatbot. Had I been using a wizard-style form I would have been more likely to at least get to the end of the interaction. In that case I would know up front that I wasn't dealing with a person, and wouldn't be put-off mid-way through.
Consider what made HAL, from the movie "2001:A Space Odyssey", so unsettling. Besides "his"
rampant paranoia and homicidal rampage, HAL was just slightly off. Uncomfortable delays in response, questions asked at peculiar times, and an always-pleasant voice all gave away that HAL might seem human, but missed by just a little. HAL dropped right into the uncanny valley and every audience member knew it from the start. No explanation or set-up was required - we all felt immediately uncomfortable with that pleasant, unceasing, red light.
When designing a customer experience, pay attention to the risk of falling into the uncanny valley. If your interaction can't be completely believable, you might want to go for the unbelievable. Most Teddy Ruxpin dolls (https://www.youtube.com/watch?v=8EshrR-xk2E)
either ended up in the closet behind other toys with its batteries removed or in the trash heap. But my niece still has her grandmother's Gunde bear and she will likely pass it on. A Gunde bear is hardly realistic, but it lasts, in part, because it isn't trying to fake it.
Teddy Ruxpin dolls are just creepy. Maybe they belong with corpses, zombies, and celebrity impersonators.
Don't try to fake your customer experience either and everyone will be more comfortable in the end. Avoid the uncanny valley. Make it real or blatantly unreal.
