The Importance of Precision in Language

It is generally accepted that for a machine to considered ‘intelligent’ and therefore classified as having ‘artificial intelligence’ it is only required to have a tiny fraction of the ‘intelligence’ of a human being. Nowhere that I have been able to find is this fraction quantified, nor is it suggested how such a thing is to be measured. It is difficult to quantify and measure something which we do not even agree what it is so I can’t be too hard on the AI crowd on that point. However I can and will continue to be hard on the use and abuse of the term and never stop calling out people who use it incorrectly, illogically, incoherently, and/or inappropriately.

In the real world, today January 30, 2018 (update: still true on July 15, 2021) the number of actually intelligent machines existing on, in, or above the planet stands at its all time high of zero, after holding steady at zero for the past 50 years or so.

Typically when we encounter something that has almost nothing in common with another thing (or just a tiny fraction in common) we do not put a modifier in front of it and call it ‘something’ that thing. For instance, one could say a carrot is a tiny fraction like a banana in that they are sort of the same shape. We do not call a carrot an artificial banana, we call it a carrot. You may object by saying but carrots and bananas are both “natural” so it seems strange that we would call either of them artificial anything. That is true so what if instead I called a carrot an imitation banana? That does not seem impossible to imagine and yet it is patently ridiculous and would be a very confusing way of naming things indeed. Why should it be any different for “intelligence”, then for the carrot?

How about instead of saying that a machine only has to have a fraction of something that is actually intelligent to qualify as artificially ‘intelligent’, we do the same thing we do for other things that are not the same as other things, and call them something else, instead of the same thing.

In the case of AI, I would suggest something like ‘modern day computing’ as right now it is simply a (slightly) different form of this using (slightly) more advanced math and statistics than have been the norm previously.

Preventing confusion in language

This is how we prevent confusion in language and is how language fundamentally works. Wittgenstein talked so much about this very point and once again the importance of his work, and the general correctness of what he had to say about language becomes evident. Our current situation with respect to ‘AI’, the fear, the mistrust, etc. so clearly demonstrates the dangers of using sloppy language when we talk about things, especially things that could be of great consequence to all of humanity. So called AI as it currently exists (it does not actually) is of zero threat and very little consequence, except perhaps to technology beat writers and Wall Street profiteers, but maybe in a few hundred years could be, or it might not. In either case I hope that whatever form modern computing aka ‘AI’ takes in the future they give it a better, more appropriate, and more precise name than what has been foisted upon it today. Mostly, this abuse of the language has happened I would suggest, because the technorati’s biggest fear is of looking ridiculous. Straight up admitting this ludicrous notion of intelligent machines is not a real thing, they would rather confuse and dissemble with a watered down definition of ‘intelligence’ for machines. For the price of their wounded egos we are stuck with a confused and fearful populace and a degradation of part of what it means to be human and what makes humanity special, our actual intelligence.