Will it make more sense to say, "I'm going to teach AI the combat art"? I don't think AI care about "style" at all. AI also doesn't care about self-cultivation, inner peace, culture study, ...
Can you teach AI not to use force and be soft?
I'm going to be specific with Jow Ga. This way I can ask questions like "Are there any martial arts similar to Jow Ga Kung fu." This will allow me to do comparisons without having to teach it about the other fighting systems.
This will also help me to identify if I really understand Jow Ga Kung Fu because I already know the answers for that question. When I teach it Kung Fu I will not go into self-Cultivation, inner peace. It will have culture study only for historical reference points. More like looking at a family tree but instead of people. Martial arts. Instead of nationality. Influences that the system may be gaining or losing.
The first thing that I have to do is teach the computer what Jow Ga Kung Fu structure looks like. I'll use this based on screenshots of video. My theory is that if I give it a series of screenshots of a movement that it will be able to start recognizing the structure of specific types of punches, kicks and other movements. I will teach it the Form and then the Application in hopes that it will learn how to translate new forms into application.
I won't worry about force right now. I will only try to teach it force if it gets all of the other things work. I don't know where any of this effort will break AI but I'm confident I can teach it a majority of the first form that I learned. Especially since I know why it was originally confusing Left and Right limbs with the photos.
It will be interesting to see what data it will collect as it tries to learn. Normally you have to tell it relevant information but you can also tell it to collect or take note of all visual data. That way the reaction in someone's face may give information that we as martial artists never thought to factor in.
I will start off with the shape of my fist
The process will always start off the same.
1. confirm what AI is "Seeing" by asking it "What do you see in this picture?" Let AI describe it so that you can make corrections
2. understand what Ai is processing by asking it "Why do you think that?"
3. Explain to AI in depth what the picture is. while addressing any errors that were made in 1 and 2.
Ai must have enough information to be able analyze new content based on the elements in the picture.. I think once I get Jow Ga Kung Fu down. Everything will be a comparison when identifying similarities. If I spar against you then there will be similarities of techniques even though the systems are the same. But in terms of information and accuracy. I know more about Jow Ga Kung Fu than other systems.
