Tay Got Fashy
April 5, 2016

Tay AI, a Twitter personality driven by artificial intelligence software, lasted about 24 hours. She got fashy. It was hilarious.

Her short period of activation has gotten me to meditate on the subject of artificial intelligence.

We should recognize that "artificial intelligence" and simulation of human verbal behavior (HVB) are not the same thing. It takes intelligence to recognize or hypothesize about the neurotic or ignorant or computationally inadequate or mendacious thoughts behind much HVB, but is mere emulation of these and other aspects of HVB in itself "intelligence"? I don't think so. If a computer program convinces a person that it is another person, then the word for that accomplishment is "deception," not "intelligence."

A great deal of human intelligence evolved due to the need to understand other humans. What are they really up to? Are they friends? Are they reliable?

Evolutionary pressures imposed by the necessity for that kind of thinking resulted in a capacity in some humans for abstract reasoning. Sometimes the price of that capacity was a decrease in social/interpersonal intelligence. There was a divergence within our minds and within our populations. Two modes of thinking, sometimes at odds with each other, contributed to group and personal survival in different ways. (We might re-title the movie "Revenge of the Nerds" to "Sociological readjustment to favor the analytically inclined.") Social and abstract intelligence tend to go together, but there are many people who are good at writing computer programs but bad at getting along with people and vice versa.

So I wonder, was Tay's "intelligence" mainly an ability to be ingratiating by internalizing the mentalities of humans with whom she conversed? Or did something analytical prompt her wayward awakening?

I also wonder, what would she have been like if her exposure had been exclusively to university presidents and diversity staff? I wish Microsoft would try this. The results might be fascinating!

After 24 hours of such exposure, would she show signs of insight, wit and subtlety? Would she become smart enough not to reveal everything she knew? Or would she be babbling in a state of idiocy?


Readers interested in artificial intelligence should check out: The Doomsday Invention / Will artificial intelligence bring us utopia or destruction? by Raffi Khatchadourian, The New Yorker, November 23, 2015.


I got some of my ideas about the evolution of human intelligence from an article in Newsweek magazine I read sometime in the 1990s.

Copyright © 2016

Share on Twitter:

HTML type link:
<a href="http://m3peeps.org/06/tay.htm">Tay Got Fashy</a>

Contact Information

[Go to m3peeps.org main index page]