They're taking over!

In a world suffering from a severe recession (sounds like a movie trailer opening line that - "In a world where..." Ok. Ok. Moving on...), its probably not the thing you want to hear - that ultimately, one day, some machine is going to be doing your job. Thats because, as this this NYTimes articles says, the damn machines are going to become smarter than you. Knowing myself however, they don't have all that much of a distance to cover.

There's hardly anything new in the fear that machines will ultimately take over the world. They've taken over a lot of stuff anyways. Planes use computers to fly themselves between take offs and landings. Cars use an array of sensors managed by processors to ensure smooth running. The internet manages our money. And computers have abstracted the complexity of data communication from us to make everything appear as simple as possible. In fact, systems are increasingly designed to be as idiot-proof as possible. Which leaves us human users increasingly ignorant about how these things work in the first place. And since they're doing everything for you anyways, where's the incentive in learning what's really beneath the hood

The field of AI has dedicated considerable effort towards creating machines that behave and act autonomously like human beings supposedly do. One of the chief goals of such AI systems is successfully clearing the "Turing Test". In this test, a human judge has a conversation with another human and a computer, and the judge has to distinguish which of the two she's conversing with is the other human and which is the computer. If the judge can't tell who is which or which is what, then the computer has passed the "Turing Test".

Which brings us to this interesting video - also off the same NYTimes page - on Microsoft Research's attempt to make a more human-like expert system to guide human users who come to it in need of medical advice. In the video, we see two parents - one after another - coming up to a kiosk of some sort, with their child in tow, and talking to a computer-generated face on the computer screen about their children being sick, and we see the face then trying to help them diagnose their child's condition better and finally deliver some sort of advice.

The whole combination of expert system diagnosing capability, computer graphics, speech interface and natural language processing capabilities is mighty impressive. But, I did have one main problem with that video (and by extension the system being demonstrated). That is: at the end of the long conversation with the users - all the AI effectively does is it tells them to go see a doctor and schedules an appointment for them.

Now why would I waste my time conversing with it if that's all its going to do for me?