Although the popular press has said less about it than they might have, IBM continues to promote its various Watson-related initiatives, which they generally speak of as cognitive computing.
To those of us who were active in Artificial Intelligence and Expert Systems in the Eighties, Watson and Cognitive Computing simply represent the latest effort to incorporate more AI into mainstream computing. For those unfamiliar with AI, it simply represents a broad program of computer innovation aimed at making computers more flexible. In the Eighties, we used to speak of AI as having three practical areas of application: Knowledge-Based Systems, Natural Language, and Robotics. All were being pursued in the Eighties, but it was generally agree, then, that Knowledge-Based Systems were the most advanced, and hence the emphasis, in the Eighties on Expert Systems – software applications that used knowledge to solve complex problems that normally required human experts. Since the Eighties there have been significant advances on all fronts.
Robotics Those who have followed this area of research, have noticed that robots are now extensively used in all areas of manufacturing. Mule-like robotics have been developed to carry loads for troops, and automobiles are increasingly able to drive themselves. Indeed, it is only a matter of time before cars come with an auto-driver mode. Similarly, commercial aircraft are able to take off, fly and land without human intervention – although we still don’t allow this yet. I wrote awhile back on the robots at Amazon that fetch items for order-processors from the warehouse. And those who have read articles on mechanical arms and other prosthetics, know that physicians are now callable of integrating very complex mechanical devices directly to the human nervous system. Robotics has certainly come of age and its hard to imagine that we won’t become more and more dependent of robotic aides of all kinds.
Knowledge-Based Systems Some readers may imagine that expert systems disappeared in the early Nineties. It’s certainly the case that interest in expert systems declined. It turned out to cost too much to maintain very intensive knowledge systems. At the same time, however, the techniques became widely used in a variety of different areas. Most advanced computer games, for example, depend on large dollops of AI. Similarly, the business rules that BPMS companies now use to document how policies are implements are all based on the rule-based techniques that were pioneered by expert systems companies in the Eighties. More to the point, we are rapidly expanding our use of rule based systems beyond applications that simply rely on business policies, and are one again talking about capturing knowledge used by human experts. All of the current work involving the use of knowledge that experts use when the deal with Case Management applications are spin-offs of expert system developments.
Natural Language In the Eighties, everyone agreed that Natural Language would be the hardest thing to accomplish. In spite of obstacles, however, Watson’s impressive win in Jeopardy, where the system was able to analyze the question, rank the right answers and choose one (knowledge-based work) and then speak the answer before the human contestants could, showed that IBM had made significant progress. The dark secret of the Jeopardy contest was that the questions were typed in before hand, and then fed, in digital form to the computer just as the human asked the questions. In other words, Watson didn’t have to first parse spoken human speech to figure out the question, but only needed to analyze a digital version of human speech. It did, however, need to identify the right answer and then generate the response.
Still, Watson represents a huge step forward, and as Moore’s law continues in effect and the power that computers can bring to the task keeps doubling, it’s reasonably clear that we will continue to make progress in both “hearing” and generating “speech.” Indeed, IBM has begun to spin out a whole range of applications that depend on Watson technology. In essence, they are mining large databases of all kinds, using what is learned to provide decision support for humans. One new application answers questions for service men, who ask questions about support available from the military as they transition back to civilian jobs. Another provides advice for patent attorneys. It a short time its obvious that IBM, using Watson technology, is going to introduce a whole new generation of user-friendly spoken interfaces for computers – interfaces that increasingly rely on knowledge-based techniques to search and summarize very large data bases to provide human users with answers to complex questions.Obviously all this AI technology can sound quite intimidating if you are unfamiliar with it, but process analysts and business analysts are ultimately concerned with how to make processes better – and this new technology makes for better processes — so its worth learning about. In essence, IBM and Watson are creating lots of new ways to assure that we can make future business processes more flexible and more responsive to employees and customers.