"Eight Marines - not a single one detected..."

Chip Cutter wrote an article that appeared in The Wall Street Journal this past week ("More Big Companies Bet They Can Still Grow Without Hiring") that caught my attention, particularly in light of some of my recent posts (see, for example, "Will we get replaced by AI?", as well as an older post, "Fewer jobs, more machines").  The "quick summary" of the article states, "Companies are increasingly aiming to maintain or reduce workforce size, anticipating AI will automate tasks and boost productivity."  It seems that not a day goes by when I don't read an article or blog post or hear a conversation about how artificial intelligence (AI) is going to replace white collar workers in the not too distant future.  For example, just this morning in the WSJ "10-Point" daily news feed, I saw an article ("Tens of Thousands of White-Collar Jobs Are Disappearing as AI Starts to Bite") that started with the ominous sentence, "The nation's largest employers have a new message for office workers: help not wanted."

Artificial intelligence has even recently passed the Turing test, a test first posed by the English mathematician Alan Turing in the late 1940's that tests a machine's ability to exhibit intelligent behavior equivalent to that of a human.  Notably, Turing originally called his test the imitation game (see the original manuscript, "Computing machinery and intelligence" that was published in the journal, Mind in 1950).  The 2014 film "The Imitation Game" starring Benedict Cumberbatch and Keira Knightley is a biographical drama about Turing, his famous test, and how a team of mathematicians helped crack the German Enigma code during World War II.

There's no question that AI is a game-changer.  I've barely scratched the surface on using AI tools such as Microsoft Copilot or ChatGPT, and I've already noticed significant benefits in productivity.  However, just like any tool, we should be cautious about how much we rely upon AI to do our work.  I've posted in the past about the potential dangers of automation and technology in general (see "Are smart phones making us dumb?", "The Glass Cage").  And I've also posted about how AI at times may not be as good as a real human (see "Artificial Intelligence?").  So, of course a recent post I saw on LinkedIn caught my attention.  I wasn't sure whether it was true or not at first.  However, after some more research, I came to the conclusion that the story was, indeed, legitimate.

Apparently the story first gained prominence after it appeared in Paul Scharre's 2023 book, Four Battlegrounds: Power in the Age of Artificial Intelligence.  The Defense Advanced Research Projects Agency (DARPA), the U.S. Department of Defense's central research and development organization, was first established in 1958 to help fund cutting-edge research that has often lead to transformative technologies, such as stealth technology, GPS, night vision, and the precursor to today's Internet (ARPANET).  The Economist magazine called DARPA "the agency that shaped the modern world."

DARPA conducted a set of field tests in 2019 to determine if a squad of eight U.S. Marines could outwit an AI security robot.  The robot was "trained" for several days to recognize humans moving in a standard fashion, such as walking or running.  By the end of the training period, the robot performed flawlessly in a series of preliminary tests.  For the actual test, Marines were instructed to use any means possible to avoid detection and touch the AI robot.

As Marines often do ("Improvise, Overcome, Adapt"), the eight Marines went off script and avoided detection.  Two Marines somersaulted for 300 meters across open ground.  Another pair hid beneath a cardboard box and simply walked up to the robot.  One Marine removed a branch from a nearby fir tree and held it in front of himself.  The AI robot saw absolutely nothing.  When asked to detect an intruder, it failed miserably!

As then DARPA Deputy Director Phil Root said, "Eight Marines - not a single one got detected.  You could hear them giggling the whole time."  The system was trained to detect predictable human movements, but it failed to detect unconventional human movements.  The Marines completely fooled the AI robot!  Max Hauptman reported on this story in 2023 (see "Marines outwitted an AI security camera by hiding in a cardboard box and pretending to be trees") and stated, "While an AI can outperform human beings in a specific task, people, as Scharre writes, have a tendency for 'mistaking performance for competence.' In other words, an AI can be very good at what it knows how to do. But the AI doesn’t know what it doesn’t know, and it also doesn’t know that it should know what it doesn’t know."

I'm not saying that this is definitive proof that we won't all get replaced by AI in the future.  I actually think that AI will continue to be a game-changer.  And the AI that was available in 2023 is very different than today's AI - and the technology is getting better every day.  But I'm also not ready to give up on us humans.  As Scharre himself wrote, "Humans tend to have a much richer understanding of the world."  So, even if a group of U.S. Air Force pilots can't beat an AI flown F-16 in a dogfight, the Marines remain undefeated against AI!  At least for now...