Wednesday, October 3, 2012

Should We Treat Military Acquisition Contracting As a Science and Use Algorithmic Decision Making?


Now then, even though I am an advocate for research technology, I don't know if I believe that everything in our society, and all the decisions we make at every level, in every business, every agency, and even in our academic and military institutions ought to be done by an artificially intelligent computer rather than a human. Whereas it is true that humans make mistakes, I would submit to you that computers can also, especially because they were programmed by a human, and because things change and there are different future parameters than there were at the time the software or algorithms were written. Let's go ahead and talk about this for second shall we?
Not long ago, I read an interesting paper by David V. Lamm from the Naval Postgraduate School "Contracting as a Science" and he discussed amongst many other things the question; "Why a Contracting Science?" He suggests that there are many ways to tackle the challenge of creating decision-making software for the military when determining which weapon system acquisitions to make, and that a decision tree could be created to pick the winners of the defense contractor bidders on every single contract - and do so more efficiently with a guarantee of no corruption, mistakes, and fewer challenges or protests of the winning bidder, as such things take time, delay projects, and cause political firestorms.
Interestingly enough, I've always been one who believes that in the future the eRepublic will be how our government is run. That we will have decision-making artificially intelligent computers managing our government, military, and academic institutions - the reason I believe this is because when a human being is given too many parameters, their ability to make the best decision starts to fall down. For an artificial intelligent computer, often the opposite is true, the more parameters given help it make the best statistical choice.
Perhaps a good example of this was when Watson, IBM's supercomputer beat the very best human Jeopardy players, and each time it gave its own best guess statistically of the prospects of that answer being correct. If you added more parameters, the statistical chance of it being right go up dramatically, as in 99.999%. Wouldn't it be nice if our government made all the right decisions? What if our military choices in the acquisitions it made for its weapons systems were always the best choice, to tackle any future adversary.

No comments:

Post a Comment