In which the author points out that often AI applications, (for example help automatons) solve problems that shouldn't exist in the first place. I happen to disagree a bit with the author and think that if AI is a pre-cursor to improved processes, that can be a win if the cost/benefit balance sheet stays positive. I'll give an example. Today there is a lot of advanced document processing and OCR going on in the grey areas between institutions. Legal filings, health records, even bills of sale, are all recorded digitally, but incompatibly. We use ML as a low-cost way to find out where the value is; once discovered, document owners will have an incentive to standardize, and the need for the AI/ML disappears. To me that's a big win.
1 comments