Friday, June 18, 2021

Artificial Intelligence in Google and elsewhere

 In the 21 January 2021 edition of the London Review of Books is the review (entitled "Insanely Complicated, Hopelessly Inadequate") by Paul Taylor* of three books:

The Promise of Artificial Intelligence: Reckoning and Judgment by Brian Cantwell Smith

Rebooting AI: Building Artificial Intelligence We can Trust by Gary Marcus and Ernest Davis

The Book of Why: The New Science of Cause and Effect by Judea pearl and Dana Mackenzie

*Paul Taylor is "a professor of health informatics at University College London."

Taylor quotes Brian Cantwell Smith to say "huge efforts are made to constrain the vocabulary used in clinicians' computer systems, but the problem goes deeper hat that.  It isn't that we can't agree on the words: it's that there aren't always well-defined concepts to which the words can be attached." 

"Smith tried to explain this by comparing a map of the islands in Georgian Bay in Ontario with an aerial photograph showing the islands along with the underwater topography.  On the map, the islands are clearly delineated; in the photography it's much harder to say where each island ends and he sea begins, or even exactly how many islands there are.  There is a difference between the world as we perceive it, divided into separate objects, and the messier reality.  We can use logic to reason about the world as described on the map, but the challenge for AI is how to build the map from the information in the photograph."

"Cantwell Smith argues that if we seem to inhabit a world that is constructed of 'discrete, well-defined mesoscale objects exemplifying properties and standing in unambiguous relations', that is an achievement of our intelligence, not a truth that can be used when engineering an artificial intelligence.  This will resonate with anyone who has tried to express seemingly straightforward concepts in sets of rules, only to be defeated by the complexity of real life."


I was once passed over for a promotion because, according to the manager, I was the only one who knew how to do the work.  So the promotion went to someone with a business degree who regularly attempted to get me to create procedures to describe how the work ought to be done.  Since I wouldn't comply, he brought in outside experts to create the procedures he wanted; so I marked them up to show how they wouldn't work.  So yes, Smith's arguments resonate with me. 

Perhaps the above is part of the reason I have never taken the Vienna Circle, especially Carnap, seriously.  Carnap for example defines everything that doesn't have its roots in empiricism as "nothing."  Godel would never say such a thing.

No comments: