Can an artificial intelligence really think?
On the old Dragnet television program, which aired in the United States from 1951 to 1959, the protagonist, Detective Sergeant Joe Friday, always says to women witnesses to crime, “Just the facts, ma’am.” This exhortation is to distinguish between what they actually observed and what they thought or felt about it. The “facts” are basically what they saw.
First off, we believe in facts that others have handed on to us, often without ways of directly verifying the data. Indeed, it would be impossible to live if we always had to verify every single fact in order to think. Even in Missouri (“the Show-Me State”), the inhabitants have to take a great deal on trust.
A fact comes to our awareness as a declarative statement, either in language or mathematical formulæ (which is also a language…). It often contains bona fides, references to those trustworthy sources who have ascertained that it is true. “I saw on the internet that So-and-so is the richest person on Earth,” one person claims to be a fact because it was found on the digital superhighway. “No, I read in the newspaper that Thus-and-such is actually the richest person on Earth,” replies the other. So who is the richest person on Earth?
Besides asking why anyone should care, resolving this difference requires finding a truly authoritative source that both can trust. In other words, an incontrovertible (verifiable) act of judgment on someone’s part.
Method in our madness
You can see that there is a method at work: getting “Just the facts, ma’am (or sir)” requires reflection. This is because facts themselves are — in fact — “artificial”. They are produced by mental processes that aim to find something true. To make a fact for oneself begins with a desire to know, continues with paying attention to the information at hand, either the interpreted data of senses or symbolic communication or both, and then asking questions. This first requires imagination as well as attentiveness, to seek insights, find clues, grasp patterns. Then when enough evidence has apparently accumulated, one formulates questions in the form, is it so? Finding a way to answer that question, the act of judgment, requires putting possible answers to the test of reality. When you have examined all the evidence, asked all the questions you can, and tested the answer to see whether all conditions have been satisfied, you arrive at a fact. Because the fact is a verifiable statement that is the result of a mental process, I say that it is “artificial”. Not in the sense of fraudulent or phony, but with the meaning of an artifice. It takes human effort to find a fact — hard work, in fact.
The origin of the word “fact” (Latin factum) shows clearly that the verifiably true statement is artificial, because people have arrived at it in ways that can be reproduced: it is “made”, ars (art) and facere (make).
Tell it out
All people who have a healthy brain can and should reproduce this method. Our ability to reproduce the reasoning process behind what is claimed to be a fact is what we have in common: it makes communication possible. Without being able to relate to one another, there would obviously be no facts to share. The mental processes we all (normally) have allow us to communicate information, in the form of facts or at least, attempts at facts (or conversely, lies covering facts, since to lie you first need a truth to hide).
Just the facts, Ms. AI
Could an “AI”, an artificial intelligence, replicate them, a computer “brain” programmed to think like we do?
That question is, of course, still an open one. However, I think we can make some remarks. First, what engages our intelligence is the desire to know. There are all kinds of desires, including wanting to know, but among them is a disinterested desire to know. Call it “curiosity”.
So what, you ask? The point is that facts begin with an urge to know, and scientific facts are the result of investigations that begin with this disinterested desire to understand: knowing something for its own sake. Can an “AI”, an artificial intelligence, do that? It would depend on its ability to be aware, first; and second, to be self-aware enough to recognize something in the field of its sense-data, or information conveyed symbolically (language again), that does not meet its expectation. Then the machine would need to be able to interrogate its memory for similar forms encountered in the past, and then imagine some correlation. In other words, it needs to be able to formulate a question on its own.
Parenthetically, the current debate about where we are going with the development of artificial intelligences does not seem to take this into account. The problem as presented by, among others, Elon Musk, is whether we should be afraid of AIs “who” have become truly self-aware, as self-aware (or more) than we humans can be. Before that becomes a real issue, the first step is to understand our own intelligence, and that means getting rid of a lot of wrongheaded “just-take-a-look-out-there” errors: knowing is not seeing. And dump a whole lot of propaganda about how “post-modernism” surpasses older ways of thinking, believing nonsense like “all truth is relative” and subtler mistakes.
We are therefore not close to building a true AI. And even if we did, it would have one fundamental difference from us humans: it would know with certitude who created it, whereas we cannot know who created us. (This would definitely influence its relations with us… but not necessarily for good…)
To build a true AI, therefore, we need first to begin with the ancient Greeks: know thyself.
 The underlying assumption was that women are “emotional” and must bring their excitability under control, at a man’s command.
 Of course, as a Christian I do believe that we are creatures and that our Origin has been revealed in Jesus of Nazareth. But this is a belief, not a fact like say, it is a fact that the sun appears to rise and set because our planet revolves on its axis.