Why I do not believe in the possibility of creating AI in the foreseeable future
Imagine that in some miraculous way a computer of an extraterrestrial civilization appeared to us. We found it, and saw that there is a program that displays “Hello, world!”.
We really liked this program, and we immediately decided to write the same one. They fought, they fought, they programmed, they programmed, they tried all the languages ​​and platforms - it doesn't work. And unfortunately, there are no USB ports on the enemy hardware, and you will not run any debuggers-decompilers.
And then we decided: let's go, so to speak, brute force. They brought a bunch of equipment and began to study alien technology. Well, there are different tips, x-ray shine. It seems to be about to finally make up a complete list of commands of the enemy processor, examine the entire architecture, and ... ')
What about "and"? Well, even compose-learn - how it will help to write "Hello world"? If we couldn’t write it with our .NETs and Vedwanballs - what will help us with the list of microinstructions of the alien processor?
And, if that, the complexity of the work of the program “Hello world” refers to the complexity of studying an unknown architecture using contactless methods, much like the complexity of the work of human intelligence to the complexity of modeling the human brain entirely.