BankThink

Flashback: Long Before Watson, Banks Explored AI

Net Spurs Banks to Look Again at Using Artificial Intelligence

By Daniel McQuillen

Artificial intelligence systems are making a comeback in the financial services industry, due in large part to the rising use of the Internet and corporate intranets.

Artificial intelligence, or AI, is a term for computer programming techniques that help systems sort data and solve highly complex problems while "learning" from the tasks.

After some false-start commercial application in the 1980s, artificial intelligence systems have gained new luster for their ability to provide tailored information to Web-using bank customers or knowledge-seeking employees.

"We've seen a resurgence-or what I call the second coming-of AI, basically because of the Web," said Allen Bonde, director of Internet computing strategies at Boston-based Yankee Group.

"If you look at what consumers are doing on the Web, they're having a dialogue through a Web site, they're shopping for things, they're doing a lot of self-service. If you can make that interface more intelligent and useful, it has a lot more value."

Artificial intelligence can involve a bevy of programming methods and computer theories, including case-based reasoning, expert systems, fuzzy logic, and neural networks. Though they typically performed well in laboratory environments, many AI technologies fell out of favor after underperforming in real-world applications.

"In the early to mid '80s the whole field was over-hyped," said Philip Klahr, vice president for customer quality with Inference Corp., a Novato, Calif.-based software company.

Now, says Mr. Klahr, with the increasing power and decreasing cost of computers and the wide-scale availability of large amounts of information, AI is again becoming an important area of investment for all companies.

Mr. Klahr said that knowledge management-the capturing and storing of a company's collective knowledge-is an area that benefits from AI.

Banks, for example, are using artificial intelligence programs to help employees sort through the immense amounts of information accessible through corporate intranets.

This information, whether used to educate employees on best practices or for improving the way a customer is served, is becoming more accessible through AI programs.

"A lot of financial institutions have huge data bases of loan histories and customers' preferences that can tell them what kind of people like to invest in what kind of products and so on," Mr. Bonde said. "But for the most part, there aren't a lot of people who understand how to get the information out of it."

The promise of these artificial intelligence technologies, he said, "is that more and more people will have access to information in a timely manner-without having to have a doctorate in computer science."

Nancy Markle, executive vice president and chief information officer at H.F. Ahmanson & Co.'s Home Savings of America, said her bank is an AI believer.

Home Savings uses artificial intelligence programs for analyzing portfolios and loans and for monitoring the bank's transactional Web site. It also plans to apply AI to its intranet.

"AI is not necessary for knowledge management, but with AI you can do it faster, easier, and more consistently," said Ms. Markle.

She also extolled AI's potential in other areas. "A new employee … can in very short order become as capable as an employee who has been there for years."

Swiss Bank Corp. also is using artificial intelligence to help employees sort through the data on its intranet.

Using CBR2 software from Inference, Swiss Bank is indexing all of its intranet-available documents and making them available via case-based reasoning, an artificial intelligence algorithm that learns to anticipate the types of information requested by different users.

Financial companies also are beginning to use AI to custom-tailor products for consumers. Swiss Bank and Washington-based American Finance and Investment, a subsidiary of Virginia First Savings Bank, are among the four financial companies now testing BrightResponse, an AI-based program produced by Novato, Calif.-based company, Brightware Inc.

BrightResponse, due to be released generally next month, uses case-based reasoning and natural language recognition to automatically answer customer E-mail.

According to John Knightly, director of corporate marketing for Brightware, "Using AI, BrightResponse can take the E-mail messages, read them, determine the intent, and either answer them automatically or route them to a sales officer if they're hot leads.

Others are planning to use intelligent software to offer personalized financial information to Web users.

Salem (Mass.) Five Cents Savings Bank plans to begin sending targeted product pitches and financial plans through its Web site in the next few weeks.

Working with Cambridge, Mass.-based software developer Vertigo Development Group, Salem Five hopes to give customers "a reason to stop by and visit the site every morning" said Michael Fitzgerald, the bank's senior vice president for marketing.

Rob Rosen, president and chief executive officer of Vertigo, warned bankers against using artificial intelligence systems that are too complex.

"You have to be really careful with the AI you use," said Mr. Rosen. "It has to be basic, because even though you might not produce the right answer all the time, you darn well better not give the wrong answer."

For reprint and licensing requests for this article, click here.
Bank technology Analytics Compliance systems Technology
MORE FROM AMERICAN BANKER