While struggling with grad school, I spent a summer thinking about how my dyslexic brain works - how I am often able to "see" complex connections in processes and systems without necessarily understanding them well (intuitive leap?). It is as if I am storing information/facts as vectors in some space where the vectors have fuzzy locations (also hard to recall) but pretty clear directions and magnitude. When I had enough vectors collected, I only followed all those "arrows" to new conclusions. Then, I had to painstakingly spent most of my time on backtracking and explaining the connections I just saw.
Doesn't that sound like Josh's Gradient Optimization AI?

Phase II: Gradient Optimization AI (November)
+ Capable of making implicit plans of moderate length
- Incapable of explaining exactly why it's doing something (everything is implicit)
~ Capable of roughly understanding interactions between plan elements
+ Capable of reacting extremely quickly to new situations
+ Capable of numerically optimizing short-term plans to a high degree, at the expense of long-term capabilities
~ Incapable of efficiently understanding parallelism

I hope I can evolve to a Bidirectional Planning AI one day