How to make artificial intelligence, and why this kind of things are so hard to create in practice.

(Wikipedia)

How to make artificial intelligence, and why this kind of things are so hard to create in practice. 

Even the best artificial intelligence solution is only the large database, where are lots of actions recorded. And when the artificial intelligence needs to do something new, the new action must be programmed by a human operator. The making of the database would take an extremely long time if we want to create a sophisticated database, where are lots of answers in multiple problems. And making those solutions is very hard to create because every single action needs to program separately. When we are thinking about a chess computer, that program would just calculate an extremely large number of movements on the chessboard.

The thing is that on the chessboard is a limited number of possible movements and the computer must just calculate them all. That makes it effective. And then the programmer must find out, what is the most effective first movement and program that movement to the program. Then it can calculate every single movement and counter movement, what the opponent can make. And of course, the terms like "mate" and the strategic object of the game must be programmed.

This is the problem to create artificial intelligence. The thing that every each action must record to the database separately is taking time. The thing with this kind of artificial intelligence is that the creation of this kind of systems is similar work with making statistics. 

Statistics is an effective tool for finding out things like the political opinions of some group. The tools for statistical analysis are simple. Only thing, what is needed is Excel or some other similar program. But collecting the data to those sheets is the thing, what takes most of the work. The same thing is with the robot, which uses artificial intelligence. 

Every single movement must record to that robot separately, and of course, there must be found some connection with the command and some action. Let's take one simple example of that thing. If the programmer wants to teach commands like "take me a newspaper". The thing seems a very easy process. But when the programmer wants to make that thing, every part of the command must teach separately. 

When the actions for the command is described the creator must describe, what is the newspaper, where it can be found, and what to do if the newspaper is not found from that particular place? The thing is that if this part of the program, what is called " the error correction" is not made, the program can be locked or stoned because the event handler cannot find matches between the input, what is taken from sensors and the action, what is stored in a database. 

The teaching of the computer of the robot must happen by using the same method what is used with children, and the operator would use the command to console, and then walk with the robot to the place, and then make what should be done. And remember that "get" is the first part of the command. 

Then the next command would be the object, what must be taken. And then the action, what must be done after that. But if the robot would not find the newspaper, it can report that thing to the controller, and if the controller would say, look it from the table. 

This kind of command means that robot would search the table, what is in kitchen and if it would not still find it, that machine would go to the living room and then it would systematically search those things from every room and every table. The thing is that the order of those rooms and the layout of the house has been programmed in the memory of the robot. 

If we think that somewhere in Universe the robots, what are equipped with artificial intelligence would start to create a new civilization, we would think that maybe they would act like ants or wasps. They would create spacecraft, which also operates as the autonomous factor, what can create other robots. 

This particular unit will have the same role with the ant queen, and it would create a new base on the planet, what it can use as the same way as ants use their nest in nature. In some vision, the robot civilization would start to create city-looking structures when they are spreading across the universe. The purpose of the city would be to act as a trap for organic civilizations. The thing what computers need is the mind of the living organisms. The problem with the computer is that they are not able to create any abstractions. So they don't have imagination at all.