The illusion of intelligence

The gap between the actual intelligence of a robot and the intelligence an observer can attribute to him is growing.

An antropomorphic body, a few low level behaviors like moving the head, blinking, synchronizing lips and speech, and people are attributing many outstanding skills to a robot partner. Those capacities never existed, but human projects his own intelligence into those behaviors: it relates head and gaze motion to what occurs in the environment, assuming that the robot has its own attentional system (although the gaze focus may be completely mostly random). 

 The very gaze of a machine can make people to assume that the robot understands what is going on in its environment, they attribute interests and intentions to him although what is going on is just motion of a few motors that can (or not) be directed toward a point in the environment.
Simple facial expressions, like smiling or nodding will create the illusion that the machine has emotions and even a model of the observer.

As the narrator of this video says with reason, the main benchmark of artificial intelligence is the reaction of a naive human... This asks again the question of what roboticists should be: illusionists or magicians? Should we create intelligent machines, or just make them appear intelligent to human beings? My answer would be that creating an illusion is easy and quick, and does not require any intelligence in the machine; however maintaining this illusion in time while the user tries to trick the machine is impossible without a proper cognition.

A long lasting illusion of intelligence requires a real implementation of intelligence. Ultimately robots should be able to trick humans to make them believe they hold a mind, but also to trick themselves...