Back in 2005 when I was first learning about “Agent Oriented Development” I was taught that agents must Perform in some Environment which it manipulates using Actuators and perceives through its Sensors (this is known as PEAS).

What reminded me about this was Kingsley Idehens 2006 blog post on the Dimensions of the Web… which is very much still a valid concept. What came to me though was that if Intelligent Agents are to be truly Autonomous on a Linked Data Web they will probably have to be mobile - and I don’t mean mobile as in developed for mobile devices, what I mean is that they should probably be capable of moving themselves from one computer to another…. otherwise they will just become some kind of “clever” web service which is stored in one place and gets its data from other web services (i.e. we revert back to Dimension 2 in Kingsleys post).

But there is a problem with this model… it sounds a bit like a virus, and something initially quite good could potentially become a bad thing with an evolutionary technique such as Genetic Programming. An intelligent agent capable of adapting to its environment through genetic programming techniques is incredibly powerful, but with great power comes great responsibility.

I have mixed feelings about Singularity philosophy, and I am particularly wary about the Singularity University, however maybe we should be thinking about the ethical/security/identity implications of Autonomous Intelligent Agents on the Linked Data Web.

Some food for thought… on this lovely Monday morning… ;-)