Can Computers Help Spot Terrorism?

ByABC News
May 21, 2002, 12:26 PM

May 22 -- Sifting through piles of intelligence data to spot a terrorist attack may require the use of artificially intelligent computer systems. But don't expect these smart systems to provide blanket protection in an uncertain world.

The obstacles are daunting. To name two of the biggest: the crush of raw information is enormous and the accuracy of A.I. systems is unknown.

"There is a ton of information out there," says Robert Steele, chief executive officer of Open Source Solutions, an intelligence analysis organization in Oakton, Va. "But the intelligence community collects less than five percent of it," says Steele.

Steele, a former intelligence officer, scoffs at the idea that computers and technology can help spot or prevent future attacks. "What is happening here is the Cold War mindset that thinks there is a technological silver bullet," he says. "There is no technical silver bullet."

But some researchers believe such a bullet exists. Norman Geddes, president and CEO of Applied Systems Intelligence, a software maker in Roswell, Ga., thinks artificial intelligence can study terrorist behavior patterns and spot trouble.

"If I'm a terrorist, I have certain goals I'm trying to achieve," says Geddes. "I need to have money, membership into an organization, and a base of operations all of which have different ways of achieving them."

Geddes says a well-designed computer program can do what humans do, and maybe even do a better job at it. "[It's] the same as a good police officer," he says. "[It] investigates leads, forms hypotheses, and narrows things down."

Almost Mimicking the Human Brain

Artificial intelligence systems use complex math routines to discover patterns and predict possible outcomes.

In general, A.I. techniques are encoded to contain a set of data about past experiences and rules that relate to that data. That allows the computer programs to "learn" and mimic the reasoning process that the human mind uses to spot patterns and trends in information.