From what I can tell, those are simply very advanced computers, not "intelligent" computers, so to speak.
Digging through all the above links, I came across the "Turing Test" (
http://en.wikipedia.org/wiki/Turing_test ). Based on what I've read, no computer has convincingly passed the Turing Test for assessing how close a computer comes to replicating human behavior. Now, it is true that computers should be able to weed out
unintelligent human behavior and thus
not precisely replicate human behavior, but at the same time, if a computer is able to weed out unintelligent human behavior, it must then know the difference between unintelligent and intelligent behavior. Therefore, if a computer is able to distinguish between human behaviors, then it must, by extension, possess the ability to know which behaviors to exhibit to pass the Turing Test. In short, a truly sentient AI system would pass the Turing Test by being intelligent enough to know precisely which behaviors to exhibit to pass the test.
From what I'm seeing, I'm inclined to vote down sentient AI as being classified as a tech in prototype form seeing as no system has yet to be developed that has convincingly demonstrated sentient thought.
However, I would absolutely encourage advanced computer networks as a tech. For example, the SEAS system you linked to is a highly predictive system that would be extremely useful for pragmatic decision making, and not just in military affairs. I can certainly see something like an advanced SEAS network being an extremely useful tech for a nation to invest in.