[alicebot-aiethics] in our own image ?

[Tim Barker]

Hi list !

I've had cause to recently think about ethical aspects of anthropomorphic 
agents. I was attending a workshop in London entitled "Animating Expressive 
Characters for Social Interactions". Various agent implementations were 
discussed and demonstrated with some very interesting and particularly 
entertaining results. However, one thought in particular related to 
behaviours expressed by some of the agents and reported of others by 
attendees which seemed to reinforce societal practices deemed 
discriminatory/offensive etc. The last thing I want to do is be prudish or 
politically correct for the sake of it but I think certain personality 
traits amongst agents may need to be considered more carefully. For 
example, one of the conference delegates reported a male agent which 
demonstrated sexist behaviour. Another reported an agent which demonstrated 
a propensity towards interrupting a conversation. Now, some of this 
behaviour may, indeed, be counterproductive. For instance, a female 
customer dealing with a website's front desk attendant may not stay around 
too long if faced with a sexist, belittling attitude. Furthermore, a 
student may decide to disengage with an instructional agent in a virtual 
learning environment who insisted on interrupting all the time ! 
Nonetheless, it may be tempting to mimic these offensive behaviours in our 
virtual agents in an attempt to create 'realistic' personifications. I just 
think though that we don't need to reflect the wealth of antisocial 
behaviour we find in the real world in the artificial ones that we're 
creating. Of course, this statement stinks of liberal utopianism, maybe, 
but I do objectively believe that we're in a position to right a few wrongs 
on this new frontier.

What do others think ?

Cheers,

Tim

Tim Barker Staffordshire U.K.
www.soc.staffs.ac.uk/~cmttimb/