Sunday, October 17, 2010

Machine Ethics in the Scientific American


Congratulations to our colleagues Michael and Susan Anderson for their article on Machine Ethics (ME) in the October issue of the Scientific American. In the article, titled Robot Be Good: A Call for Ethical Autonomous Machines, they introduce both ME and their recent work programming ethical principles in Nao, a humanoid robot. Nao, pictured to the right, was developed by the French company Aldebaran Robotics.
Nao is capable of finding and walking toward a patient who needs to be reminded to take a medication, bringing the medication to the patient, interacting using natural-language, and notifying an overseer by e-mail when necessary. The robot receives initial input from the overseer (who typically would be a physician), including: what time to take a medication, the maximum amount of harm that could occur if this medication is not tak- en, how long it would take for this maximum harm to occur, the maximum amount of expected good to be derived from taking this medication, and how long it would take for this benefit to be lost. From this input, the robot calculates its levels of duty satisfaction or violation for each of the three duties and takes different actions depending on how those levels change over time. It issues a reminder when the levels of duty satisfaction and violation have reached the point where, according to its ethical principle, reminding is preferable to not reminding. The robot notifies the overseer only when it gets to the point that the patient could be harmed, or could lose considerable benefit, from not taking the medication.

Those familiar with the Anderson's work will appreciate that Nao is the first robotic implementation of work they did on EthEl.