There’s nothing inherently wrong with the technology. However, doing ML well isn’t really about the ML as much as it’s about getting everything leading up to t

There’s nothing inherently wrong with the technology. However, doing ML well isn’t really about the ML as much as it’s about getting everything leading up to the ML right. That’s the difference between value-add applications of ML and an expensive mess. IBM didn’t understand that and hence the mess they’re in now with Watson.