Three robots built by roboticists at the Ransselaer Polytechnic Institute in New York were put through the classic 'wise men puzzle' test of self-awareness. One of them passed, but as they’re coded the same it could be said that all three passed.
In the test, two of the robots were prevented from talking. All three were then asked which one was still able to speak. All of the robots were supposed to say, “I don’t know,” but of course only one succeeds.
When it hears its own voice and not any of the others, it then understands it was not silenced and says, “Sorry, I know now!"
This sounds like a pretty easy task for humans, but it’s a bit more complex for robots. The robot must first listen to and understand the question asked, hear and identify its own voice, and then connect that back to the original question to understand that it hadn’t been silenced.
This test is a take on the wise man puzzle in which a fictional king is choosing a new advisor and gathers the three wisest people in the land. He places a hat on each of their heads and promises them that at least one of them is wearing a blue hat. He also says that the contest is fair to all three of them.
Each wise person can see all of the other hats, but not their own. And they cannot speak to one another. To become the new advisor, the person must stand and correctly announce the colour of their own hat. Can you work it out?
Logical puzzles like this that require an element of self-awareness are essential in building robots that can understand their role in society.
Selmer Bringsjord’s work will be showcased at the RO-MAN conference in Japan, which runs from 31st August to 4th September 2015.