Sounds good, right?  But once you take a minute to think about what this really means, you soon realize it is totally dependent on trial and error.  Take an extreme example.  You ask a computer if the U.S. should make a preemptive military strike to disrupt North Korea’s nuclear program.  If the computer is programmed to “believe” depriving Kim Jung Un of nuclear weapons is the “prime directive,” its first response is “yes.”  If Kim Jung Un does not respond to the attack and gives up his nuclear arsenal and development program, the computer will make the same recommendation in the next instance.  But what if North Korea launches a response resulting in 100,000 South Korean casualties?  Based on that experience, the computer might “learn” from its unfortunate recommendation and come to a different conclusion the next time (though it is a little late).

Which distinguishes machine learning from “deep thinking,” a form of artificial intelligence dependent on a massive neural network which approximates that of the human brain.  Humans do not have to wait to learn from experience.  They can anticipate outcomes and raise questions, such as, “Is the potential loss of live worth the outcome?”

As a student of political science, I would argue military policy since the nation’s founding has been an exercise in machine learning.  From the Revolutionary War up until Korea, experience suggested America could achieve its national and international goals through armed conflict.  We won our independence in 1783.  We preserved that independence in 1812.  We liberated Texas in 1836.  We avenged the “attack on the USS Maine” in 1898.  We joined with allies twice to win the “War to End All Wars” in 1918 and to defeat the Axis Powers in 1945.  The underlying algorithm, simply stated, “We fight!  We win!” Machine learning ended with Korea.  A computer, absent any ideological bias, would have responded, “Wait!  Experience tells me we don’t always win.  Sometimes we end up in stalemate.”  The Vietnam experience would force the computer to further challenge the “we always win” algorithm.

The one instance in the past quarter century when we should have heeded machine learning, the first and second Gulf Wars, we failed to listen.  In 1991, we should have learned the United States is on solid ground when it asks other nations to join us in deterring one global player (Iraq) from attacking another sovereign nation (Kuwait).  But that algorithm did not hold water during the second Gulf War when the United States became the aggressor.

Deep thinking, drawing on the cognitive powers of a complex neural network, would have appreciated the difference between conventional warfare and insurgencies.  Between fighting nations and fighting dispersed, borderless extremist movements.  But as Arlo Guthrie once said, “That’s not what I came here to talk about.”  Because understanding whether a White House is steeped in machine thinking instead of deep thinking tells us a lot about how it will respond to a crisis with potentially catastrophic consequences.

Which brings us to the behavior by Donald Trump and his staff over the past two months.  This White House exhibits a tendency to stick to machine thinking when deep thinking would serve it much better.  And the best evidence is how it handles the continuing departure of executive staff.  I am not suggesting that firings and resignations are equivalent to declaring war or deporting Dreamers.  I am only pointing out a decision making process which bodes ill for good policy and problem solving.

On December 13, 2017, former Apprentice contestant and White House staffer Omarosa was fired and forcibly removed from the West Wing.  She is now in a verbal war with the Trump White House following negative comments she shared about her time in Washington on her latest reality television appearance (CBS’ Celebrity Big Brother).  Fortunately for Trump, to this day nobody knows what Omarosa did for the administration and it is highly unlikely she had access to any discussions or documents which might be of interest to Robert Mueller.

But the machine thinking mechanism in the White House processed the input and came up with the following conclusion.  “Treating a former employee badly will cause them to turn against you.”  So if we’re going to fire someone, we better be nice to them,  especially if they have had access to sensitive information.

Enter Rob Porter, a serial spouse abuser.  The computer says, “Better be nice to him.”  And Trump was. Enter Hope Hicks.  Trump’s initial, and for once correct, reaction is she had acted improperly by drafting a defense of the object of her latest office romance.  But the computer said, “Remember, Omarosa showed us hell STILL has no wrath like a woman scorned.”  And of course, Hicks was on board Air Force One when Trump drafted the cover-up press release concerning Junior’s and Jared’s June, 2016 meeting with the Russians in Trump Tower.  “Better be nice to her.”  Within hours, Trump called Hicks, “…absolutely fantastic…smart, very talented and respected by all.”

So add Porter and Hicks to the list including Vladimir Putin and Stormy Daniels who reinforce a machine thinking algorithm which predicts “blackmail works.”  We always knew Trump and his posse were driven by “artificial intelligence,” we just didn’t know which kind.  Now we do.  Instead of state of the art “deep thinking” which is so desperately needed in a era of complex problems and nuanced opportunity, Trump is dependent on 20th century technology to go along with his 19th century vision of America.

For what it’s worth.
Dr. ESP