The puzzle is based on a fictional king, who in order to choose his next advisor, invites three of the land’s wisest men to court. There, he puts either a blue or white hat on each of their heads and tells them that the first to stand up and correctly guess the colour of their hat will become his advisor. Similarly, Selmer Bringsjord of the Rensselaer Polytechnic Institute prevented two of three off-the-shelf Nao robots from speaking before asking them which one was still able to speak. The robot who could still speak said “I don’t know” before hearing the sound of his own voice and becoming aware of his ability, when he added “Sorry, I know now!”.
This is the first robot to pass such a logical puzzle. While this is a simple test for humans, it is not for robots, which are programmed to do what we tell them to. Such a test requires the robot to listen to and understand the question, hear their voice and recognise it as their own, before connecting it to the question and coming to a conclusion.
Logical puzzles requiring self-awareness are essential in building robots that can understand their place in the world. By passing tests such as the wise man puzzle, it is hoped that robots will be able to build up a group of human-like abilities that become useful when combined.
Bringsjord will demonstrate the wise man puzzle at RO-MAN conference, which takes place from 31st of August to 4th of September 2015 in Kobe, Japan.