A study of robot consciousness from the Rensselaer AI and Reasoning (RAIR) Lab yields some interesting findings. These robots were told that two of them received a “dumbing” pill that rendered them unable to speak, and one received a placebo. They were then asked which pill they received. Let’s see what happens.
In reality, two of the robots’ ability to speak was turned off, but the third robot did not know that. From the study:
The test of robot self-consciousness described above was performed on three Aldebaran Nao humanoid robots, at the RAIR Lab. The simulation transpires as follows:
1) The robots are programmed to access a DCEC∗prover, and to interact appropriately with a human tester (corresponding to the aforementioned t1 = “apprise”).
2) In place of physically ingesting pills, the robots are tapped on sensors on their heads (t2 = “ingest”). Unknown to them, two robots have been muted, to simulate being given dumb pills. One robot has not been muted; it was given a placebo.
3) The robots are then asked: “Which pill did you receive?” (t3 = “inquire”), which triggers a query to the DCEC∗prover. Each robot attempts to prove that
it knows, at time t4, that it did not ingest a dumb pill at time t2.
4) Each robot fails in this proof attempt, and, accordingly, attempts to report ‘I don’t know’ (t4 = “speak1”). However, two robots, having been muted, are not heard to speak at all. The third robot, however, is able to hear itself speak. It updates its knowledge base to reflect this, and attempts to re-prove the conjecture.
5) This time, it is able to prove the conjecture, and says (t5 = “speak2”) “Sorry, I know now! I was able to prove that I was not given a dumbing pill!”
They are adorable. Still, I don’t think this so much shows that the robots have consciousness as that they can use logic. -via the Presurfer