Mar 292013

LSI-CAP_site-header_480x122A computer dubbed Watson is on the loose, and that means it’s silly season. Again. Watson, you’ll recall, is a computer that IBM, its proud creator, touts as displaying a new level of artificial intelligence. IBM invented the machine specifically to play the TV quiz show game Jeopardy, which it did quite well at in a 2011 contest with two men who’d won big on the show itself. Since then, and with some fanfare, Watson has been put to use as an advisor to doctors and nurses, helping some to diagnose ailments and others to decide on treatments for patients with lung cancer. Speculation is rife about what other applications this kind of machine might be useful for. IBM execs have mentioned insurance, legal research, telecom, and banking. Clearly, like every big computer company, IBM wants to be perceived as a leader in technology, and equally important, in helping customers apply technology in profitable ways. If nothing else, Watson is a big help to IBM in that regard, as is the firm’s DeepBlue computer, which plays grandmaster-level chess. Just this past week, IBM fellow Grady Booch, a big name in software development circles, made headlines here (and here, too) talking about how it’s “inevitable” that computers will one day be sentient — having feelings, that is. Reportedly, he made this comment in a talk at a West Coast computer history museum. (Unfortunately, no video or transcript of his comments is available on the web.) Assuming Booch actually does believe that computers are destined to acquire human feelings, he is not alone. Inventor and author Ray Kurzweil has amassed a sizeable following with his argument that computing and bio-engineering technologies are now advancing at such a rate that machine intelligence will exceed that of humans by 2030, or perhaps earlier. The human mind, Kurzweil states, is in essence just another computer program, potentially describable in binary code that would, one day, be available for copying, editing, storing, and loading back into a human brain for execution. Somehow, I doubt this very much. Perhaps one day computers will have feelings of some sort, or at least the ability to make us believe that. Already, there are computers that seem able to understand natural language and create new sentences based on some digitized description of the world and human ideas. Watson, for instance, has been fed several terabytes of text documents, including the entire contents of Wikipedia, which it is programmed to search at great speed and, to a fashion, reason about.