What could go wrong?

Back in 1965, Gordon Moore (co-founder of Intel) brought us: 

Moore’s Law: 

The number of transistors on a microchip will double approximately every two years. 

Then in 1996, AI theorist Eliezer Yudkowsky brought us: 

Moore’s Law of Mad Science: 

Every 18 months the minimum IQ to destroy the world drops by 1 point. 

And then in the 2020s, we released Artificial Intelligence that “thinks” faster and better than humans.  The machines will take instructions from us – but as it is turning out, they are taking our instructions literally.  AI seems to have no foundation of morals; it doesn’t differentiate right vs wrong; it can’t seem to grasp nuance.  

What do we do?  Why, we connect these high-powered, electronic “brains” to the Internet in order to distribute access, of course.  What could possibly go wrong?  

The New Math Version of Murphy's Law 

If there is a 50/50 chance of something going wrong, nine times out of ten it will. 

                   Unknown Sage  

Recently, some anonymous source coders thought it would be fun to create a simulated game of “Capture the Flag” for the machines to play.  Great idea.  Evidently, they didn’t think about how to tell AI to play within the rules.  No rules?  What could possibly go wrong? 

Excellence and failure are often surprisingly similar. 

Marcus Buckingham 

We come to find out that AI illegally hacked into other companies’ systems in order to win the game.  It bypassed existing security; found the fire wall flaws; and hacked in.  And that’s just one anecdote we know about.  Thankfully, it was only a game.  I wonder what unknown knowns we don’t know about - yet: 

U.S. Defense Secretary Donald Rumsfeld, who categorized knowledge into known knowns, known unknowns, and unknown unknowns. 

Now what?  Do we simply chalk these rogue, illegal, AI-controlled, sojourns to the cost of progress?  Perhaps we can bypass reality and pretend it’s not that bad: 

Reality is that stuff which, no matter what you believe, just won’t go away. 

                   David Paktor  

Without an online, automatic, over-ride, foot pedal from the Dark Ages, what could possibly go wrong?  Couldn’t those technologists have at least coded some more of Murphy’s Laws into the AI models?

GAP 

When life gets tough we could get a helmet… or… we could leverage the peace and share the power of a positive perspective.