By Christmas 2010, billions of autonomous software agents are expected to be roaming the Web year-round. Researchers believe these shopbots will transform e-commerce, shopping for goods and services for people and companies. Consumers already use rudimentary bots on comparison-shopping sites that query multiple retailers and retrieve prices and other information for users. Some bots attend on-line auctions for buyers and bid on their behalf.
Researchers are now designing bots to handle complicated pricing algorithms and complex automated tasks, such as stock market trading, travel arranging, and purchasing and procurement between companies. Because these intelligent agents will be acting as economic decision makers, researchers want to know more about bot behaviour before they’re turned loose on the Web. Jeff Kephart is manager of the Agents and Emergent Phenomena Group at IBM Corp.’s Thomas J. Watson Research Center in Hawthorne, N.Y. He and his colleagues study the collective behaviour of models of large populations of software agents. He says, “If you’re going to introduce a new species into the world’s ecosystem, it’s prudent to understand the consequences.”
Researchers found that bots tend to incite price wars by compulsively undercutting each other. Kephart is optimistic that researchers can write out bad behaviour and construct an entire economy of super agents that support one another – and human beings.
“A decade from now, almost every person and company will be represented by one of these agents,” he says. “The possibilities are really mind-boggling.”
For more information on IBM’s bot research, visit www.research.ibm.com/infoecon.