Uncanny Valley

There is a psychological effect in which the more a thing is recognized as being human the more trustworthy or relatable it is. There is a brief gap of cognitive dissonance when we can’t decide anymore if something is human or not. This gap is the uncanny valley and we generally don’t trust things if they are to be placed in there.


The uncanny valley finds many expressions in the interaction with voice assistants. First, we have uncanny communication by encountering human voices but robotic speech patterns. The voices of current implements are near perfect, indistinguishable from real humans, but the way they communicate is robotic or scripted.

Another problem that is often encountered is that the devices go off without users intending them to do so; activation without interaction. I have more than one funny and sometimes downwards creepy anecdote to tell around this. These seemingly random activations are often not perceived as quirky, like with a pet, but alienating.

The last issue is that the voice assistants build up a ghostly presence. This perception arises from the fact, that voice assistants are always-on devices. They are always listening in on their environment, waiting for a wake word, upon hearing it being unlocked and available for further interaction. Knowing this made my participants feel a presence in the room that led them to alter their behaviors and how and what they said.