The article is skeptical and superficial, but if I were to assemble a research program to create real-world, operates-out-in-the-wild intelligence, my one page executive overview would look a lot like the Air Force requirements. This set of problems is going to be nibbled away a bit at a time. Some of them may be horrendously difficult and some may be low hanging fruit.
I think that this will be open research and not filed away in some secret warehouse a la Raiders of the Lost Ark.
So this is kind of like the study of how humans think ... or their "cognition", say. You could call it a ... cognitive science.
I only mention this because the phrase "cognitive science" isn't mentioned once in the linked article or the call for proposals. Cognitive scientists have been working on the problem of finding the "algorithms of thought" for decades. Look at Marr's Vision, or Randy Gallistel's body of work (in particular his new book Memory and the computational brain) for examples.
Noah Shachtman gets a downmod for no research. We have some idea of what the components of the mind are now. We can relate all of the branches of mathematics and science to specific areas of function with an evolutionary basis.
I think that this will be open research and not filed away in some secret warehouse a la Raiders of the Lost Ark.