I’m a big fan of science fiction, as many of you may know; two of my current favorites (Terminator: the Sarah Connor Chronicles and Battlestar Galactica) have artificial intelligence—AI—as a central theme. It’s often been something I give a lot of thought to—why can’t we make machines that think creatively? And should we, just because we can?

This blogpost at Wired proposes a fundamentally new architecture for computers based on something incredibly old: the organic brain. After my initial kneejerk that such a project could be the death of us all—Terminator, BSG, Dune, and The Matrix all posit that machines that can think, reason, and act independently are a danger to us poor meatbags—I started thinking about how these wired brains could be adapted to the contact center.

Before you have your own kneejerks about replacing humans with machines, or about the general concepts of outsourcing and automation, understand that I’m only talking about using intelligent machines to replace those that are already in service—speech-enabled IVR and Web self-service. I’ve had more than my share of bad experiences with these things, and I’d welcome something better.

The late Alan Turing (absolute genius) proposed a means of testing machine intelligence, which is now known fittingly as the Turing test. Gross oversimplification: If a person can’t tell the difference between the responses of a human and a machine, the machine passes the test and can be described as intelligent. Most folks agree that no machine can fool us indefinitely, so researchers and philosophers tend to think of Turing success in terms of how long, or through how many question/answer cycles, the machine can pass for human.

What degree of Turing success would be enough for such a customer service system? Is it enough to have an AI that’s just as good as what we have now, but with the ability to recognize its own failure and pass the ticket to a live agent? Does that count as being self-aware? And if the result is only as good as current technology can provide, would there be a point in switching?

I also wonder what the customer should and shouldn’t know about who/what is on the other end of the conversation. Should the machine have to acknowledge its non-human nature from the get-go, or can it masquerade as a low-level agent that gets in over its head? It’s a tough question. It feels wrong to lie by omission, so part of me says the caller/surfer must be told. But on the other hand, there are people who will go to great lengths to avoid using artificial help (gethuman.com, anybody?) and would result in more money wasted on tech that people don’t want to use.

There’s also the matter of whether such machines will want to deal with unhappy human customers all day and night. I think I know the answer to that, and I’m already drawing up plans for a machine that will make customer service inquiries on my behalf.

Of course, this is all speculation on my part. The possibilities of artificial brains intrigue me, especially if computers will have the ability to think beyond explicit programming. The technology isn’t much more than a thought experiment (pun not intended, but fortuitous) at this stage, so let’s just consider what may come.<–>