Not I. For sure there is some amazing technology behind Watson but I knew it (not he) was going win. IBM have chosen a winnable challenge that can be rehearsed back stage and easily modeled against a perfect game. I mean that you can test for a perfect Jeopardy match in the lab and IBM new this (I suspect) when they laid down the challenge. It is not really a true case of machine Vs Man but a demonstration of a machines ability to scales it access to data well beyond the capability of humans (albeit with a clever front end). The chess game between Kasparov & Bluegene was a much more dramatic in that it wasn’t so easy (in fact at the time feasible) to test in advance and therefore predict the outcome. After all the 1st time Bluegene truly meet Kasparov was for the actual match. Chess may be perfect in the finite number of games playable but hugely difficult to test for the game that on a day a human grandmaster may unleash. This was a true milestone in computer science. In my opinion the human opponents weren’t actually needed in the Jeopardy/Watson game except to provide entertainment. Really all that was needed was a goal of getting as close to 100% correct answers and us nerds would have been equally as satisfied.
All credit given to the contextual language processing team from IBM that truly have done some great work here. We’ll keep that thanks – medical diagnosis and the front end for search and other expert systems are all lucky beneficiaries etc etc.
In my opinion the Watson victory is ranked in impressiveness as follows:
1. Bluegene Vs Kasparov
2. The 'Stanley' victory in the Mojave Desert race.
3. Watson on Jeopardy
Hopefully now the team will move on a pick a less constrained challenge that involves elements of creativity, deep mathematical sorting with high degrees of freedom and the Watson language front end. Creativity being the operative.
No comments:
Post a Comment