1. Neither of us is complacent. If something is broken - do something about it.
2. Both of us are musicians who write and perform politically driven music.
3. We both push the boundaries. No society advances without someone constantly provoking it. People hate it but it needs to be done.
4. I guess there may be a slight resemblance between my picture and some from his site but you judge. I personally thought I was more like Billy Idol from Gen-X. Met the guy once - we do look alike.
Do we look alike?
So what the hell does this have to do with artificial intelligence and Adobe LiveCycle. Artificial Intelligence is a huge waste of time and will never work. What is promising is the concept of Computational Intelligence or getting computers to aspects of mimic intelligence based on our expectations. There are two very important and almost always neglected aspects of CI. One is the lack of context. Context is everything, especially for inference. Look at the Sex Pistols putting out Never Mind the Bullocks made a lot of people at the time very upset for a number of reasons. It was insulting to the queen of England and talked about real things we all think. Keep in mind this perspective, the Queen of England once hated Rock and Roll, but has since knighted several rock musicians. In the context of rock’s invention, it was not accepted however over time, it became an acceptable art form (WFT that really means). Looking back now, the Pistols got an entire generation to sit up and say "something’s not working here and were pissed". In the right context, I would nominate Johnny and the boys for giving an entire generation the message that they better think for themselves and things can be changed. It changed my life. When I had the chance to work for the United Nations and be disruptive and tell people who were full of crap that they were full of crap. I did it and I am proud of that.
The point is that Context is everything and most approaches to artificial intelligence seem to be rather static than dynamic. Most of the AI research to date in the field of Ontologies and semantics seems to take a hard coded approach. This is not how humans think.
To add to the problem, there is one other important thing that humans do well that computers cannot do as well. This is excellently summed up in David Luckham who I consider a genius in every sense of the word. Definitely on par with Johnny. Simply stated, it is the ability to detect two events, recognize the context in which they occur and understand the causality relationship between them For example, if you came home and saw that your spouses car was gone, that is one event. If you then saw a guy carrying a TV around the corner, that in itself is another event. Now put the two events together and the causality may point to a situation where they are stealing your TV because the house is empty and they helped themselves. Easy for a human, not so easy for a programmer to capture this model in a generalized sense so it can be reused over multiple examples.
To make Computational Intelligence work, one would require a model for Complex Event Programing, an inference engine using something like the Blackboard patterns with a hypothesis limiter on it to negate the exponential hypothesis problem, a context ontology to layer over top of whatever semantic reasoning one might employ, and a large enough source of events that be used to feed it all. The latter is very important since event isolation would lead to an incomplete set of events to mine for the inference component and probably miss key things.
This is where Adobe LiveCycle enters. LiveCycle, as a platform, has several key places where events are captured and stored in a manner that they can be audited later. The next generation of the LiveCycle platform and the blades that plug into it carry a much more complex series of audit trails than the last version.
Here is a hypothetical situation where a phishing attack is starting. A mass email goes out and tells customers of a bank to use their forms to change their password because of a security breach. If you use LiveCycle Forms, served from a LiveCycle Form Manager, you will generate an auditable event each time one of your customers grabs the form. If your customers account is then accesses and a transfer form is filled out, that will also generate a second event. You now have two events that have a relationship to one thing - the unique account. This should be a relatively easy pattern to catch if you are using LiveCycle as a platform over your IT infrastructure. If you were to use multiple disparate technologies, it could also be caught however this may take more hard work to account for different Document models and mine the events at the same level of granularity.
LiveCycle Policy Server will be able to make events accessible in the LC 8 release and via the SDK, some events are available now. This should be useful to companies who are serious about thwarting criminal behaviour linked to people using their IT infrastructure.
Back to Johnny Rotten now. Things have to change in the world. The pistols and PIL changed me and now I am feeding back change into the world. People need to start thinking about this stuff and start finding ways to fix it. Come to think of it, I guess I am not too changed from my previous punk days. The picture on the left was taken in July of 2006 ;-)