February 16, 2011, 7:59 PM — Like many, I have watched IBM's Watson computer compete on the answer and question show Jeopardy, wondering what IBM has created. During the course of the show, my mind regularly wandered off, wondering whether this event has broader implications. Having been briefed on Watson prior to the show's airing, it is clear that Watson is not HAL (see 2001, A Space Odyssey), and yet, within the confines of the Jeopardy challenge, IBM's innovators have created a skilled Jeopardy competitor second to none. One wonders whether Watson would have achieved the same result if it only had 5 racks of computers instead of 10 (since Watson seemed quicker to respond) or whether it would have fared as well without much of its pre-show training.
From a practical standpoint, the obvious question from a futurist's perspective would be "what's next?" If one looks back on Deep Blue (chess champion) and Blue Gene (biological modeling powerhouse), IBM suggests that the platform has the potential to do great things, something IBM dedicated significant resources to demonstrate in pilot projects. That said, IBM efforts have not translated significantly into successful commercial offerings or broadly adopted transformative approaches to problem solving. Can Watson prove to be the exception?
I thought it was notable that Watson was built on COTS components that could be purchased by a prospective customer tomorrow. It is clear that IBM has invested heavily in a custom software application that whose specific goal was to successfully navigate the Jeopardy ecosystem. This application cannot be readily modified to address either societal or industry problems, but could be somewhat helpful in better translating human communication, idiomatic speech, and double entendres in the course of improving information searching. Beyond that, it is likely that Watson's achievements are, at most, a baby step along Ray Kurzweil and Time Magazine's projection of a computer surpassing human intelligence by 2045.
So what should IBM expect to gain financially from its investment in Watson? Is it different from Deep Blue or Blue Gene? In this case, maybe yes.