Polytropy

Somebody else can amuse us or frighten us. They might put us to sleep with an injection, or perhaps keep us awake. They cannot make us rational. They cannot make us think. Thinking is up to us.

We cannot then expect to make a rational machine. Thinking has to come naturally – not in the sense that it can be studied by physics, but that it cannot be manufactured. Artificial intelligence is a contradiction in terms.

I have been writing a lot about AI lately. I have not been doing it here, although some of the writing may end up here. I did blog about AI last summer, in “Artificial Language” and “Gödel and AI.” There is always more to say.

In a video that I watched as a child in the National Air and Space Museum, Julia Child cooked up a “primordial soup” (see Sarah Zielinski, “Julia Child and the Primordial Soup”, Smithsonian, September 22, 2010). If intelligence could evolve from such a soup, then perhaps it could arise in a computer. In that case, there would be no reason to expect the intelligence to do what we wanted.

Whether what is called AI today will actually turn out to be useful, I don’t know. Some people say it is already useful; however, the cost of the usefulness is not yet clear. As far as I know, the AI companies are not turning the profit they want, and their spokepeople deserve as much credence as the current president of the United States of America.

Some people say they have conversations with AI. But then, people talk with their pets and their plants.

I did deign to try out AI on the grammatical puzzle that I brought up in Appendix 1 of “Biological History.” In the sentence

ἁπλῆν μὲν πρᾶξιν ἧς γινομένης ὥσπερ ὥρισται συνεχοῦς καὶ μιᾶς ἄνευ περιπετείας ἢ ἀναγνωρισμοῦ ἡ μετάβασις γίνεται

from Aristotle’s Poetics (§ X.2, 1452^a14), does ἧς γινομένης ὥσπερ ὥρισται συνεχοῦς καὶ μιᾶς constitute a genitive absolute clause whose subject is a relative pronoun? An AI told me it didn’t, along with a lot else, which I did not find enlightening. Later, a human Greek scholar straightened me out: the relative pronoun ἧς is genitive because it has ἡ μετάβασις as object, so that the whole sentence has the following analysis:

λέγω δὲ ἁπλῆν μὲν	|	I call simple
πρᾶξιν ἧς	|	the action of which,
γινομένης	|	– [it] coming about
ὥσπερ ὥρισται	|	as has been defined
συνεχοῦς καὶ μιᾶς	|	continuous and one –
ἄνευ περιπετείας	|	without reversal
ἢ ἀναγνωρισμοῦ	|	or recognition,
ἡ μετάβασις γίνεται.	|	the change comes about.

Perhaps some day AI will be a better Greek scholar, but what would the point be?

I decided to write the present post while reading Chapter 3, “Nature and Action,” of Collingwood’s “Principles of History.” I have written about this posthumously published work before, especially in the post “Biological History,” already mentioned, and in “Freedom,” back in 2014. I could write a lot more, but for now I’ll just report a note I found in my copy of the book. I probably wrote it last year, when a friend had been reading his copy.

What I wrote was, “Nobody else can make you be rational. Ergo computers are not!” This was opposite the page with the following paragraph:

The discovery that the men whose actions he studies are in this sense free is a discovery which every historian makes as soon as he arrives at a scientific mastery of his own subject. When that happens, the historian discovers his own freedom: that is, he discovers the autonomous character of historical thought, its power to solve its own problems for itself by its own methods. He discovers how unnecessary it is and how impossible it is for him, as historian, to hand these problems over for solution to natural science; he discovers that in his capacity as historian he both can and must solve them for himself. It is simultaneously with this discovery of his own freedom as historian, that he discovers the freedom of man as an historical agent. Historical thought, thought about rational activity, is free from the domination of natural science, and rational activity is free from the domination of nature.

I wrote in “Law and History” about a natural scientist (Peter Turchin) who thought historians could hand over their problems to him. Now there are people who want to hand their problems over to a computer. Is that any different from following the recommendation, “Let Go and Let God”?