E-Mail
AI: Thinking Outside the Box
by Lynne Ellyn, Fellow, Cutter Business Technology Council
Artificial intelligence (my definition): AI is technology based on the study of human cognition and problem-solving capabilities. Examples of such human cognitive skills include speech recognition, decision making, visual recognition, problem solving, deductive and inductive reasoning, and signal processing. Some of the technologies that came into being from AI research include multitasking/processing computers, robotics, object orientation, expert systems, rules engines, neural networks, fuzzy logic, fractal databases, and pattern matching.
Tesler's Theorem: "AI is whatever hasn't been done yet" [1].
From Google's search engine to call center voice recognition systems, applications based on AI technology are changing the face of business. Many such AI-based applications are taken for granted by the users and consumers, who expect such technology will keep getting smarter. The drumbeat for innovation and never-ending competitive pressure has kept the people trained and skilled in AI techniques employed and productive despite media and popular opinion that AI "failed."
While product companies have pushed forward in the use of technologies and approaches borrowed from AI research, inhouse IT departments have largely been clueless about the possibilities for innovation based on this technology. As IT departments have standardized and outsourced, most have lost any ability to produce breakthrough business results through development based on "advanced technologies." With the current focus in IT limited to purchased applications or what can be produced using Java or .NET, it is little wonder that I receive an endless stream of articles and conference invitations focused on a strident demand that IT innovate now or become irrelevant. While it should be obvious that limiting tools and approaches used also limits solutions, most IT departments have not yet expanded their toolsets or methods to embrace potentially more valuable technologies.
A quick look at some of the current effective uses of various AI technologies points the way toward what can be accomplished and can get IT executives thinking outside the limitations of today's common approaches. Here are some examples of exciting applications of technologies formerly known as artificial intelligence.
According to the US National Science Foundation (NSF), "www.BeattheTraffic.com [is] a tool that tells commuters how long they can expect to sit in their cars, which shortcuts will get them home faster that day, and even the best time to leave the home or office" [2]. The application, which is now deployed in 45 cities, reportedly uses a combination of a statistical database that is collecting real-time data from 14,000 sensors and a flexible routing engine. Scheduling and routing and constraint-based reasoning systems were a primary research activity for AI researchers in the 1970s and 1980s. Companies still struggle with the best ways to optimize field service workers, respond to changing field conditions, and optimize travel patterns. Savvy IT departments with a need to innovate in scheduling or transportation systems would do well to hire (or train) people in the optimization methods and quantitative approaches first used by the AI community; the business will benefit from the innovative thinking that is sure to occur.
As many of the original AI researchers grow older, they may be pleased to find that AI technology is being deployed to help provide the care that the aging baby boomers need. Japan is leading the way in research, development, and marketing of robots that can provide nursing care for elderly or impaired people. There are now robots that bathe patients. Amazing personal mobility-aid robots with advanced navigation abilities are being introduced that can help the elderly navigate roads and sidewalks and avoid obstacles. Japan, which has a critical nursing shortage, is looking forward to the introduction of robots that know when to turn bedridden patients to help prevent bed sores. Why should IT departments be interested in such advanced robots? As IT is asked to innovate in support of business results, pairing such advanced robotics with manufacturing and supply chain systems is clearly an opportunity to bring new value to the corporate landscape. The potential for robots that can sense and respond to the environment or the people in the environment opens up incredible possibilities for the marriage of information systems and robotics in customer care, logistics, and the handling of documents.
Security is yet another place where the deployment of AI techniques and technology is being applied effectively. According to an NSF press release, the fight against the use of the Internet to coordinate terrorist activity is being given new oomph thanks to the Dark Web project [3]. This project is using the most advanced pattern-matching techniques in a coordinated way to find terrorist communications that have been deliberately obfuscated across many Web sites. IT groups at financial institutions could readily employ such techniques to develop better ways of finding potential security breaches. Any business seeking to protect its intellectual property could use such techniques to find information leaks or corporate espionage activities. Pattern matching is such a powerful way of approaching even mundane problems that every IT department needs some expertise in this approach. Back in the early 1990s when I managed an AI group at Chrysler Corporation, we successfully used pattern matching to find duplicate freight bills and prevent the payment of those bills. Somewhere in my files I still have a letter from the auditor saying that we had saved the company $12 million in just six months of using pattern matching to prevent duplicate payments. Finding patterns in complex data is a challenge in virtually every company and thus a real opportunity for IT innovation.
In the stock market applications space, neural networks, genetic programming, fuzzy logic, pattern matching, and quantitative analyses are routinely married to create the "secret sauce" of the trading floor. On the energy trading floor, which is supported by my IT department, I introduced these advanced technologies over five years ago. In our company, these applications help predict where demand for electricity will be in the future (two hours to months). This enables the traders to buy and sell more profitably. Today, the trading function is dependent on the edge given by the application of these technologies. Companies that need to understand consumer behavior and make predictions about market demand can certainly improve their performance with the skillful use of these AI techniques. This is a mostly untapped innovation opportunity and IT can lead the way.
Pattern matching, robotics, fuzzy logic, fractal analysis, voice and speech recognition, vision recognition, inferencing, constraint management, and complex scheduling are just a few of the technologies that IT can use to enable extraordinary business results, if only they can develop a vision of the possibilities. In time, this must happen in order to gain the competitive advantage. Today's crop of common technology, while useful, is not equipping IT to be the innovation leader that the business so desperately needs. Now is the time for CIOs who want to lead their companies in innovation to acquire the AI skills and tools that will make it possible to get breakthrough results. Just don't call it AI.
I welcome your comments on this issue of the Cutter Edge and encourage you to send your insights on the market in general to me at lellyn@cutter.com.
-- Lynne Ellyn, Fellow, Cutter Business Technology Council
Notes
1. See Wikipedia's "AI effect."
2. See NSF's press release "Powerful Tool Crunches Commutes."
3. See NSF's press release "Scientists Use the 'Dark Web' to Snag Extremists and Terrorists Online."