Robots show a degree of self-awareness by passing “wise-men puzzle”

In case you’re like me (prior to writing this piece) and have no idea what the “wise-men puzzle” is, let me just provide some background on this test of self-awareness:  a king is looking for a new wise man for counsel, and so he calls three of the wisest men in the kingdom to his quarters. While all four are in a room, the king places a hat on each of the men’s heads from behind so they cannot see it. He then tells them that each hat is either blue or white, that the contest is being conducted fairly, and that the first of the men to deduce the color of his own hat wins. 

Now, in order for the contest to be conducted fairly, all three men would have to have the same colored hat. That being the case, the first man to notice the color of the hats on the other two men and announce that his hat is the same color, wins the wise man contest. 

Roboticists at the Rensselaer Polytechnic Institute used this puzzle as a means for determining if they had advanced the programming behind their Nao robots to the point that they were self-aware. Here’s how:

Instead of hats, the team programmed the three humanoid bots to “believe” that two of them had been given a “dumbing pill” that caused them to be mute; however, they were not informed as to which of them had not received the pill. 

In actuality, all the team did was mute two of the robots by pressing a button atop their heads. 

The three robots were then asked which of them had not received the dumbing pill. All three tried to respond with an answer of “I don’t know” but only one was able to actually vocalize its response. In doing so, it realized it was the one not muted, and changed its answer, declaring that it was the one that had not received the dumbing pill. 

Now granted, the exercise described above is a bit simplistic, but it does demonstrate progress in terms of developing self-awareness within robots. 

Per the group behind this development, incrementally adding abilities like that which is described here, will lead to robots with more useful attributes. Selmer Bringsjord, who represented the team at a press junction where these robots were demonstrated, explained they are not necessarily concerned about questions of consciousness; rather, they want to build robots that are able to perform tasks that might be considered examples of conscience behavior. 

For more information into their study, read the group’s published report “Moral Reasoning & Decision-Making: ONR: MURI/Moral Dilemmas”.