As I attended the UIST conference this past week, it became more clear to me than ever that there are two distinct kinds of AI. One kind is trying to be your best friend. The other is one you never meet, but every day it makes your world better behind the scenes.
I understand that there are many billions of dollars to make by selling consumers a powerful drug in the form of a make-believe best friend. Eventually people will catch on, and everyone will understand that like any powerful innovation — think automobiles and opioids — chatbots will need to be regulated. Appropriate regulations will follow.
But meanwhile, there is a different kind of AI everywhere, below the surface. The fact that you can buy inexpensive sensors which used to cost hundreds of dollars or more, that your mail is efficiently and accurately sorted, that your monthly utility bill is much less than it might otherwise be, all of this is due to machine learning working away for your benefit, in places where you never see it.
As we rightly come to grips with the epidemic of make-believe humans unleashed on an unprepared public, we might take a moment to appreciate the distinction. It’s not AI itself that is the problem — in fact it’s great. The problem is people who are profiting off particular varieties of it in arguably unethical and questionable ways.