What are the methods of implementing artificial intelligence?

There are 2 different ways in which artificial intelligence can be realized on a computer:

One is to use traditional programming techniques to make the system appear intelligent, regardless of whether or not the methods used are the same as those used by human or animal organisms. This approach is called the ENGINEERING APPROACH, and it has already made results in some fields, such as text recognition and computer chess.

The other is the simulation method (MODELING APPROACH), it is not only to see the effect, but also requires that the realization of the method is also the same or similar to the method used by human or biological organisms.

Genetic algorithms (GENERIC ALGORITHM, referred to as GA) and artificial neural networks (ARTIFICIAL NEURAL NETWORK, referred to as ANN) are of the latter type. While genetic algorithms simulate the genetic-evolutionary mechanisms of human beings or organisms, artificial neural networks simulate the activity of nerve cells in the human or animal brain. Both approaches can usually be used in order to obtain the same intelligent results. With the former approach, it is necessary to specify the program logic in detail manually, which is still convenient if the game is simple. If the game is complex, the number of characters and activity space increases, the corresponding logic will be very complex (exponentially growing), and manual programming will be very tedious and prone to errors. And once an error is made, the original program must be modified, recompiled, debugged, and finally a new version is provided to the user or a new patch is provided, which is very troublesome. When using the latter method, the programmer has to design an intelligent system (a module) for each role to control, this intelligent system (module) does not know anything at the beginning, like a newborn baby, but it is able to learn, and can gradually adapt to the environment to cope with a variety of complex situations. This system often makes mistakes at the beginning, but it can learn from its mistakes, and it may be corrected the next time it runs, or at least it will not be wrong forever, and it will not be necessary to release a new version or patch it. Utilizing this approach to AI requires the programmer to have a biological way of thinking, and is a little harder to get started with. But once it is in the door, it can be widely used. Because this method of programming does not require detailed specification of the character's activity patterns, applying it to complex problems usually requires less effort than the previous method.