Thursday, September 30, 2010

Can robots be Three Laws safe?

I'm with about 25 people in a hotel in the New Forest to talk about the ethical, legal and societal issues around robotics. We are a diverse crew: a core of robotics and AI folk, richly complemented by academics in psychology, law, ethics, philosophy, culture, performance and art history. This joint EPSRC/AHRC workshop was an outcome of a discussion on robot ethics at the EPSRC Societal Issues Panel in November 2009. (See also my post The Ethical Roboticist.)

Of course in any discussion about robot ethics it is inevitable that Asimov's Three Laws of Robotics will come up and, I must admit, I've always insisted that they have no value whatsoever. They were, after all, a fictional device for creating stories with dramatic moral ambiguities - not a serious attempt to draw up a moral code of robots. Today I've been forced to revise that opinion. Amazingly we have succeeded in drafting a new set of five 'laws', not for robots themselves but for designers and operators of robots. (You can't have laws for robots because they are not persons - or at least not for the foreseeable future.)

I can't post them here just yet - a joint statement needs to be drafted and agreed first. But to answer the question in the title of this post - no, robots can't be Three Laws Safe, but they quite possibly could be Five Laws Compliant.


Postscript: here is a much better description of the workshop on Lilian Edwards' excellent blog.

2 comments:

  1. Check out "The Ethical Robot":

    http://www.youtube.com/watch?v=ZLdvCDFriTQ

    ReplyDelete
  2. Thank you Michael for the link to your excellent video - I was not aware of your (and Susan Anderson's) project aimed at programming robots with ethical behaviours.

    In fact the focus of the workshop that I reported above was not so much the ethical behaviour of robots, but of roboticists.

    Programming robots to behave ethically is (I'm sure you'll agree) technically very demanding, and something that will require years of research before it can become integral to real-world robots. Our view was that a more immediate and pressing concern is ethical guidelines for designers, manufacturers and sellers (or operators) of robots. Something we feel especially important for robots that might be designed to work with children or vulnerable people.

    ReplyDelete