I just returned from Rensselaer Polytechnic Institute in Troy where I watched IBM’s Watson computer compete against humans in a game of Jeopardy. The huge auditorium of the EMPAC building was full of people (mostly students) interested in watching the competition on the big screen and listening to the panel of experts, alumni from RPI, who were responsible for designing and programming Watson. One of the panelists said that even though supercomputers like this were great at gathering information and answering questions in the Jeopardy format, they are still a long way from being able to do the same thing with audio or visual clues. I can attest to that personally since I use voice software everyday and the programmers still have not figured out how to come up with the right homonym in context. It got me thinking about when supercomputers would be smart enough visually to identify plants as well and as fast as a trained botanist. There are projects out there that identify plants visually, one using mainly tree leaves, and others that would read the DNA bar code of plants to identify them. Some groups of plants might be easy to identify but when you get down to grasses, sedges, willows, hawthorns and other difficult groups you might have a real challenge on your hands.
One of the panel members was asked why compete on Jeopardy. He explained that after they finished beating chess world-champion Garry Kasparov with the supercomputer Deep Blue they needed a new challenge. One night after discussing this they saw everyone rush out to a bar to see Ken Jennings compete on Jeopardy. They figured that would be the next big challenge; to show how far computers have come with their information-gathering power. So far the humans are no match for Watson (although Watson confused Toronto as an American city in final Jeopardy). I think one of the next big challenges for IBM would be to visually identify a wide variety of plants faster than our best botanists (with a supercomputer called Linnaeus?). I think that might take awhile. – Steve Young