> That is like saying "why would you use math to figure out the area of that rectangle when you can just guess randomly until you find a fitting number".
> Both solutions work but one is intelligent.
That's the same distinction between an old-school program that tells the machine exact instructions "go forward 3 meters, turn left, go 1 meter", vs. artificial intelligence - building a robot that can adapt to it's environment. Why make it adaptable when you can program in precise instructions to begin with?
The entire field of AI is based on the idea that adaptable is better. We know of about 8.7 million species, from ones that live in boiling acid to blue whales and humans. If you wanted to make million variations on something, is it smarter to explicitly design one, then the next, then another, nine million times, or is it more intelligent to design one or ten which have the ability to morph into whatever variation is required?
You're assuming hard-coding specific values is the smart way. In almost al cases, hard-coding is the dumb way to do things.