Documenting the Coming Singularity

Monday, March 12, 2007

AI is Back, Baby!

Actually, artificial intelligence didn't go anywhere. It's been quietly growing more and more powerful, and more and more unobtrusive. You're benefiting from AI every day and don't even know it.

AI has only seemed to disappear, due to a phenomenon described by Ray Kurzweil (hope you're not tired of me quoting his work) he calls the "technology hype cycle," which occurs during technological paradigm shifts, such as the railroad frenzy of the nineteenth century and the Internet and telecommunications booms and busts more recently.

It "typically starts with a period of unrealistic expectations based on a lack of understanding of all the enabling factors required. Although utilization of the new paradigm does increase exponentially, early growth is slow until the knee of the exponential-growth curve is realized. While the widespread expectations for revolutionary change are accurate, they are incorrectly timed. When the prospects do not quickly pan out, a period of disillusionment sets in. Nevertheless exponential growth continues unabated, and years later a more mature and more realistic transformation does occur."

The fact that this has happened with AI will be obvious to anyone old enough to remember the hype, when the term "AI" was seemingly on every tongue. Then it just sort of faded into oblivion, so much so that most of us think the technology failed or was simply abandoned. Not so. The fact is, AI has continued to be developed and used in a growing number of practical applications.

Another phenomenon has also contributed to the seeming disappearance of AI. It seems that "as soon as an AI technique works, it's no longer considered AI and is spun off into its own field."

But what is AI? According to Elaine Rich (a computer scientist),

"AI is the study of techniques for solving exponentially hard problems in polynomial time by exploiting knowledge about the problem domain."

There, did that help? I didn't think so. In AI's toolkit are expert systems, Bayesian nets, Markov models, neural nets, genetic algorithms and recursive search. Let's consider some of the successful applications of AI.

  • Software programs routinely diagnose electrocardiograms using pattern recognition applied to ECG recordings. Every major drug company is using AI programs to develop new therapeutic drugs through pattern recognition and data mining.
  • Pattern-recognition software systems are in use in autonomous weapons like cruise missiles to guide them to their targets.
  • Google and other search engines use AI-based statistical learning methods and logical inference to rank links.
  • Companies use AI sytems to work out optimal logistics for their complex supply chains.
  • Airlines use AI to send landing planes to available gates.

There are many more systems in use today that do things that benefit you and me and which are completely operating in the background so that we are unaware of their operation and even their existence. They solve problems that humans are not capable of dealing with due to the overwhelming number of variables involved.

Admittedly, all of these applications fall into a category called "narrow" AI. Kurzweil estimates that we will see "strong" AI, that is, AI that exceeds human intelligence, by the mid-2020s.

If you'd like to know more about Kurzweil's work, visit, or pick up a copy of Singularity by clicking on the graphic below.