Note on Writing AI characters

It is quite unusual for me to write futuristic or science fiction works but a recent challenge I set myself – to tell a particular story three times in a single short play, each at different historical junctures – led me to set a piece in the future. As a consistent character in the story is a police officer, I got the notion (not original, I own) that law enforcement in a possible future may be the responsibility of AI machines.

I do not know very much about AI, so it behoved that as well as setting my imagination to work, I read a little into the science of AI. I am not unfamiliar with fiction around the area, especially in film, having viewed the likes of 2001: A Space Odyssey, Blade Runner, Westworld and Robocop a number of times. I chose to read Artificial Intelligence: A Beginner's Guide by BlayWhitby (Oneworld Publications, 2008). It’s a simple book and suits my purposes perfectly, as I don’t wish to stage the science, just have some idea that there can be a science behind what I have happening on stage.

I have become especially intrigued by a major problem in AI, in that it has proved particularly difficult, if not impossible, for programmers to get their AI machines to distinguish between objects – the robots cannot tell a mobile phone from a glasses case, for example. This set me thinking about how this might be solved (and I am not in the least bit a scientist!) – but surely the answer might well come from looking at the problem from the other way around? The cell phone and the glasses case will tell the machine what they are. This can be done very simply by having all objects microchipped. Machines entering any space will scan the space and collect data from the microchips in the space, so that every object will be distinguishable from every other and identifiable to the machine. The microchip might well include details of the dimension of the object and even dimension of each part within, so that objects packed or piled together will be distinguishable. The microchip could also include ownership information. There might be some objects –like match sticks or fresh food – which a machine might encounter without a microchip; these would be dealt with in the same way that computerised counters deal with “unexpected items in the bagging area” – people would have to help out the machine. People themselves would be microchipped alongside objects…

The initial microchipping will involve a large-scale project to ensure that all spaces are machine-friendly. Will people agree to this whole-scale adaptation of their environment for the benefit of AI machines? I think about the way my xBox kinect works and the way in which my partner’s nephews and nieces are happy to adapt where they stand in the room for the benefit of playing a game; humans have always been willing to adapt their behaviour for the benefit of technology as long as they believe they're getting a pay-off…

In solving, at least for myself, one of the issues of AI, I can see how my machines in my play might negotiate environments. I am not sure I need to explain this within the play, but I can happily write them moving through the world without wondering how they know what’s around them.