The ambiguous term "artificial intelligence"
Lately it seems everyone is talking about the successes of artificial intelligence (AI), often meaning by this yet another achievement in the field of generative neural networks. And as usual, some make money on the hype, others crave the appearance of Sarah Connor a real strong AI (which will conquer the world in the future), and someone is just once again raping a journalist.
And there are very few who, when mentioning AI, try to explain what they themselves understand by the term “artificial intelligence”. After all, it is one thing to write about “a field of tasks of the AI category”, and quite another to endow an ordinary computer algorithm with at least the rudiments of reason.
I have long wanted to write that the very etymology of the established phrase “artificial intelligence” is ambiguous and can take different meanings depending on the context, since it denotes a concept that may differ from the one the author tries to put into it. And to understand what is wrong with the phrase “artificial intelligence”, let us break it down into its components:
Dissecting the term “intelligence”
Unfortunately, there is no exact generally accepted notion of the term “intelligence”. Even man himself (who by definition possesses intelligence) does not know what it is (this is like a nuclear physicist — when atoms study atoms). If we combine dozens of definitions, the wiki will tell us that intelligence is:
a quality of the psyche, consisting of the ability to be aware of new situations, the ability to learn and remember on the basis of experience, to understand and apply abstract concepts, and to use one’s knowledge to control the environment of a human.
Notice, there is a reference to the human as the source of this very “intelligence”, and this seems to imply that everything that is not human is not intelligence? Then the term “intelligence” is explicitly connected with the human and human activity and is a kind of “definition by contradiction”.
But if we treat the interpretation of the term “intelligence” more carefully, and pay attention to the fact that the application of abstract concepts and the use of one’s knowledge to control the environment is intended not simply for the sake of change, but for change for the sake of achieving a goal, then the meaning of the term “intelligence” begins to play in completely different colors!
If there is a goal and there is a will to achieve it, then “intelligence” is not just the possibility to create something or creatively apply existing knowledge (a dumb computer can do this too after receiving the corresponding command); intelligence is one’s own desire to do something to search for a goal, to achieve it, and this is inherent only in a living being.
And it seems to me that such an interpretation of the term intelligence (not only the presence of the ability to solve creative tasks, but also the presence of one’s own intention to set these tasks for oneself) is more correct and puts everything in its place.
Therefore, it is not enough to have the ability to do something; one must first have one’s own desire to do this “something”. In other words, the term intelligence is connected not only and not so much with creative abilities, but with the impulses to use these abilities and to apply them independently.
Dissecting the term “artificial”
The term “artificial” is much simpler. “Artificial” is something made not by nature (i.e. not in a natural way), but with the help of rational activity. But should we clarify the definition of “rational activity of a human”? Usually, in defining the term “artificial”, a similar reference is contained:
- created or carried out by a human like or instead of a natural, genuine one
- connected with the conscious activity of a human
In other words, the term “artificial” seems to assume that everything that is not done by a human is natural (of nature), whether it be an animal’s burrow or an anthill, a beaver dam or a bird’s nest. In other words, the term “artificial” seems to presuppose in advance the authorship of its creator, which, in the context of the term “intelligence”, becomes practically oil-upon-oil.
So let us ignore the reference to the rational activity of a human, and assume that made not by nature could be made not only by a human.
Moreover, the term artificial may also have another meaning, more suitable for this case, which means “not alive”. An artificial arm, an artificial tooth, etc. That is, this is also not a natural (not of nature) object, but in a slightly different sense, i.e. not simply created artificially, but precisely not alive.
Dissecting the phrase “artificial intelligence”
And since the term “artificial” can be interpreted as “not alive”, then the phrase “artificial intelligence”, in other words “not alive intelligence”, will be interpreted a bit differently!
An artificial (not alive) object does not need its own desire to do something, to seek goals in life or to achieve them. After all, this is inherent only in a living being. And since the not alive has no goals of its own, then for the term “intelligence” the contradiction that was imposed by the original definition of this term through a “definition by contradiction” is removed.
After all, a not alive computer program really does not have, and cannot have, its own desires for the application of its existing creative abilities (if we assume that it has any). It also has no consciousness of its own (since it is by definition a not alive object), but nothing prevents an external command from launching a process of creativity, as a result of which the result of its work will manifest itself (all the more so, a human can also be given an “external” command to launch a creative process).
Will the result of the work of such a program be the result of the work of artificial intelligence? If we talk about intelligence as a property of a living organism, then unambiguously no, but we are talking about the “artificial”, i.e. “not alive” version of this term!
P.S.
When I started writing this article, I was furiously indignant at everyone hyping this topic, since I sincerely believed that the term AI is used incorrectly. However, unexpectedly for myself, the rubber duck effect worked for me (Rubber duck debugging), when I explained to myself that my rejection of this term was connected with the fact that the term “intelligence” was firmly associated in my mind with personality, free will, i.e. with a living object, whereas it should be perceived as “not alive”, after which everything fell into place ;-)
P.P.S.
All the more so because it is very easy to verify such an interpretation of the terms. It is enough to compare the results of the work of an AI program and a human from the point of view of assessing the “intelligence” of the one who solves it. After all, a brilliant discovery will not be more or less brilliant if it is made specifically by a human. And to check the result, ordinary blind testing is enough (when the evaluators do not know whether they are evaluating a product of human or artificial intelligence).