The way Watson and the Deep Neural Networks (DNNs) work are very complex. It is almost a black box like Skinner proposed for behavior. I don’t think there is a way of checking how a conclusion is reached. You could delete every reference to a behavior, but because people lie cheat and steal...not everyone and not every time, of course the computer would use the absence or presence of information as part of the trained network. DNNs just got out of the toy-box, in language recognition. With reliable data from more than a few million in the customer database and a few GIGABYTES on each person, the 2% outliers will be the ones in need. The real answer
will become:
We know what is the best treatment for you. And here are your options...by what and who it costs.
Then comes the harder part of what we will do when we find out that info.
DK
I hope not as I really like CVS.
Walgreen’s can take a hike IMO
When they say “about 30 minutes” it really means about 60 minutes.
YMMV
Whoever says they have to use it just for medical care? They can have a separate secret one for “enemies of the regime” or “politically incorrect”
and a (ahem) special tag which feeds those names into the rest of the process (for the “real” no moar care flags) midstream.
Who’s gonna reverse engineer, or DNN the whole encapsulating process?
(Because, like you said, “science!”)