AI research experimentation should be conducted by the military branches.

The reports I'm getting is that the stuff being developed by the industry is too complicated; and what's more, too expensive.

This possibly could be true--to keep out competitors.  What we need is intensive military experimentation with an eye to open up the area to simplistic solutions and algorithms.

Youths, going through a military training program in AI technology could then take it to the industry, but it's got to be greatly simplified.

Can you imagine a group of African kids learning stuff that only a tech with a PhD in computers can now do?

See what I mean.  Take a year or two, but it's got to be simplified greatly.  What is this about "deep learning?"  Nonsense.

My opinion.