Giggling US marines outsmart AI with gaming technique

Eight US Marines recently cracked the code
02 February 2023

As Artificial Intelligence systems take over our lives - A new challenge has emerged - ”Could a person beat an AI in a fight? And if so, how?"

Award-winning defense expert Paul Scharre in his book Four Battlegrounds: Power in the Age of Artificial Intelligence details how the marines cracked the code.

The United States - Defense Advanced Research Projects Agency (DARPA), worked with  soldiers to refine an AI system's ability to detect people. Scharre writes that after six days of training the algorithm by having soldiers walk around,  the engineers on the project flipped the script, pitting the soldiers against the AI by placing the robot in the middle of a traffic circle and tasking the soldiers with approaching it undetected.

Knowing they can’t beat an AI system with traditional camouflage, the marines deployed childish antics and clever tricks outside of the AI’s programming regime. 

And guess what? All eight marines beat the AI system. How?

The key, they discovered, is to act like video game characters. Some of the tactics included - 2 soldiers somersaulting 300 metres,  a soldier disguised as a tree, and 2 soldiers hid under a cardboard box and were not detected even though it is reported they giggled the whole time.  

The military has long used video games to train soldiers and advertise itself to young people, for which it has come under intense scrutiny. 

- CyberBeat


About CyberBeat

CyberBeat is a grassroots initiative from a team of producers and subject matter experts, driven out of frustration at the lack of media coverage, responding to an urgent need to provide a clear, concise, informative and educational approach to the growing fields of Cybersecurity and Digital Privacy.

Contact CyberBeat

If you have a story of interest, a comment, a concern or if you'd just like to say Hi, please contact us

Terms & Policies >>


We couldn't do this without the support of our sponsors and contributors.