Ben Goertzel reports on the Workshop on Machine Consciousness at Nokia Research in Helsinki over the last couple of days. Will Browne was also there, so I am hoping to get some more in depth insights into the discussions that went on. I would have loved to have been there myself.
Ben mentions his theory about abstraction – this, and the description of Harri Pirkola's ideas on the topic remind me of the plans I had in 2001/2002 to link a ANN's (A) inputs to the weights of another (B) and train it to produce a compressed version of them (by having a symmetrical network which tapers to a pinch point and back out, and which learns by comparing output directly to input).
I really must get on with it and do that some time. I also want to use a similar technique for producing a 'network overseer'. To be honest the idea for this came about as a way of trying to produce a simulation of some 'weird hierarchy' of networks to drive 'demon' characters in a game I keep meaning to write, but I think it may have some more serious applications too.
Ben also mentions he is a panpsychist. I have trouble with panpsychism, mainly for the reason Ben outlines – I hold that 'consciousness' is a reflexive, cognitive process. On the other hand, my spiritual beliefs fit quite well with panpsychism. So maybe intellectually I am trying to hold a form of dualist view – that consciousness (the sort that some people feel makes us different to other animals) is a physical phenomenon, but that there might just be a 'spirit' side too. I do not see the 'spirit' as being essential to the working of the consciousness, however. My work on collaboration in communities leads me to think there is some form of emergent consciousness at that level too. Whilst this has the problem of starting to sound like Universal Consciousness, it does appear likely that tightly knit groups of individuals form communities which have 'spirit' or 'a life of their own'. Our legal system reinforces this, by treating companies and organizations as individuals in many ways, and I do wonder whether this is a reflection of an empirical observation or whether the 'folk' view conditions us to see communities in this way.
Anyway, all good stuff, and I look forward to hearing more about it!
Tagged your site as artificial at iLinkShare!
Comments
Linking ANN inputs to weights
But what would it achieve?