The head of Army Cyber Command said last week that the service was constantly observing and refining human oversight of artificial intelligence, with the possibility of releasing computer agents to do some work by themselves.
Speaking at the Association of the United States Army Conference in Washington, Lt. Gen. Christopher Eubank described an iterative human-in-the-loop process in which military personnel learn in real time how to balance giving AI agents autonomy and exercising control over them.
“It’s occurring daily and weekly inside of the headquarters as we watch agents learn and grow, and then determine what we should let them do on their own versus have a human in the loop,” Eubank said. “All work right now is safeguarded human-in-the-loop as we continue to learn across this very important emergent technology.”
Military personnel currently watch AI agents perform routine tasks, with a governance board within Army Cyber Command making decisions on whether that agent should be allowed to continue performing a task without human intervention or whether it should be allowed to work unsupervised.
During its current human-in-the-loop iteration, every machine AI team checked in with the human team daily and was not operating alone.
Every AI agent was also assigned the same work role as a human, so Army Cyber command trained those agents to be masters of their craft in the same way the Army did with the humans.
As the service worked through the iterative process, the uptick in speed that AI ushered in was drastic.
In two days, an AI agent could complete work that took an analytics support officer six months, Eubank said.
Army Cyber Command used artificial intelligence to operate, secure, defend and attack networks, and make sure that technology works hand in hand with the human workforce, according to Eubank.
Artificial intelligence also enabled the command to move faster than it had ever moved and sift through mountains of data at machine speed, which allowed humans to focus on solving critical problems.
