In the 1980s, there was a surge of Artificial Intelligence. Thousands of systems were built and programmed to replicate their human counter-part(s). However, in the mid to late 1990s we experienced a “winter of discontent”. Is history destined to repeat itself? Bruce Porter, Professor and Chairman of the Department of Computer Sciences at the University of Texas, thinks not.
Porter recently spoke at IBM’s World of Watson conference in Brooklyn, NY about the future of cognitive computing and its relationship with AI. He explained that in order to prevent what went wrong in the future, we must first look back on our past. In 1955 a man by the name of John McCarthy defined the idea of building intelligent machines. The first good example of this idea arrived around 1960 and it was a machine that could play checkers against a human opponent. All this machine did was replicate what a professional checker player would do when presented with a specific move from an opponent. You could say that early AI pioneers were truly psychologists at heart in that they wanted to directly compare human cognition with computer cognition.
The problem with the example above and others related is that the first wave of AI in the 1980s focused on what humans did well, and aimed to replicate that with a machine. Porter explains that at the time there was actually a competing strand of technology to AI and it was IA, which is Intelligence Amplification.
Porter explains that there are two reasons for the winter of discontent. The first is simple, expert systems simply became too expensive and timely to build. The second reason is that expert systems were “idiots”; that is, they were amazing at the task they were programmed for, but useless for anything else. They did things that doctors and lawyers already knew how to do. The builders of these systems never asked the doctors and lawyers what it was they needed help with. The problems solved were for the wrong reasons.
Here’s an example: AI (Artificial Intelligence) is like walking into an accounting firm, asking who the best accountant is, and then studying what makes that accountant good and programming those skills into a computer. IA (Intelligence Amplification) is like walking into that same firm, asking the best accountant “What is it you need to do better?”, and then creating applications that assist the accountant in his/her weak points.
IBM Watson is commonly described as an AI system, when in fact you can argue it is IA. Since IA is in its infancy, Porter predicts that the coming advances are rooted in natural dialog with the system and a focus on discovery instead of search. Instead of searching for what cancer looks like, Watson can one day discover what cancer looks like…and thus the cognitive era will have begun.
