It's time to take a stand: against the hype around artificial intelligence, but for the idea behind it.
Artificial Intelligence - A brief terminology
The word 'Intelligence' can also be interpreted as 'Enlightenment'. In contrast to 'intelligence', however, the term 'enlightenment' does not offer the same projection surface and, moreover, provides a much more realistic picture of the current state of Artificial Intelligence. They are simply more or less complex algorithms that allow a given probing of states from sensor data and thereby enable the decision for a concrete action (among several conceivable ones). This interpretation is also familiar from military reconnaissance, the aim of which is to obtain reliable information about the intentions, distribution and behavior of enemy forces.
The Hype Around Artificial Intelligence
A lance must be broken: Against the hype about artificial intelligence, but for the idea behind it. On the one hand, the term 'artificial intelligence' offers a remarkable projection surface for formulating wishes, expectations and ideas about what can be expected from this area of technical development and what possibilities can be derived from it for users. On the other hand, however, this projection surface must not be confused with what is currently technically feasible.
An Object Offers More Than Just Optical Properties
A good example is given by the frequently used aspect of Artificial Intelligence, which refers to object recognition. However, instead of attributing to an artificial neural network the ability to recognize or even comprehend its environment, artificial object recognition can rather be compared to a nocturnal scouting party that estimates the number of enemies and their equipment. In neither case is it a matter of identifying the persons or objects, but merely of recognizing essential characteristics. This comparison is also more apt with regard to the transmission of information: It is not the general himself who creeps through the grass at night. Rather, the reconnaissance team will report its findings to him.
Object recognition by Artificial Intelligence is about comparable qualities: That is, the recognition of essential properties of the object, but not the recognition of the object itself. For example, neural networks currently name the class to which an object belongs, such as a car or a screwdriver. Neural networks also recognize subclasses, such as car types or the quality state of the screwdriver, if needed. However, corresponding information is not stored after this recognition process and is assigned to the specific object in such a way that it becomes identifiable: "I recognize you, didn't you have winter tires on last time?
The Road to Human Intelligence is Still Long
Therefore, the pure object recognition remains at first a mechanical and purely enlightening process, which lacks the essential moment to intelligence: The object in the context of its environment, its history, its purpose. It is still up to the general to deduce the consequences from all the facts, an appropriate study and his experience.
What we want to see is called 'Artificial Intelligence'. What we are witnessing today is rather 'Artificial Enlightenment'. But as we know, the age of enlightenment is followed by modernity, so that some development of 'Artificial Intelligence' can be expected.