I'm not convinced that its that elementary but even if I'm wrong I think the cost, even of just adding and testing some new "pretty elementary
" software and hardware would no way be offest by the potential gain. How many games has Watson played and how many times has it happened that Watson gave an answer that a human player already gave and was rejected? I think that's happened once. And even in this case we don't know that if Watson had used the knowledge of the human answer to prune the set of possible answers that it would have come up with the correct one. Adding new bells and whistles that are barely if ever useful is feature creep and its the kind of thing that causes lots of software projects to fail. The IBM developers clearly made the right choice IMO.
About Smarter IdeasRead the Article at HuffingtonPost
No comments:
Post a Comment