- A recent US Air Force test involved artificial intelligence in kill chain decisions.
- The service said the AI helped speed up decisions and reduce the cognitive load on operators.
- It's one example of how the US military is working with AI in combat-related roles.
The US Air Force used artificial intelligence to speed up its targeting decisions in a recent exercise.
The goal was to test how AI could improve the accuracy and timeliness of operator responses, the Air Force said, in order to have a more automated kill chain in a potential future fight.
The four-day Experiment 3 exercise, held last month, was a stress test of future warfare technologies and tactics in a realistic combat scenario.
"By replicating this novel methodology, we allowed tactical-level participants to explore how tools and workflows could shape kill chain execution for the future of warfighting," Lt. Col. Wesley Schultz, the director of operations for the 805th Combat Training Squadron/Shadow Operations Center-Nellis, said.
A kill chain is a military decision process that guides how forces detect a threat, zero in on it, act, and evaluate the outcome. The US military is actively looking to evolve its kill chain through automation, AI, data-driven command and control decisions, and interconnected sensors, surveillance, and reconnaissance tools.
During the training, operators used AI software to accelerate decision-making and targeting processes. It was designed to speed up those processes and "reduce the cognitive load" on human operators, the service said in a release.
Reducing the cognitive load, or mental effort, of warfighting personnel has become a priority, especially in recent years with the rise of AI. US military leaders and officials believe that AI can assist humans in assessing massive amounts of battlefield data to help make clearer decisions more quickly.
The AI used was developed under the Maven Smart System, a larger US military AI initiative. During testing, the assessments made only by operators were compared to those made with AI recommendations.
The testing was intended to explore the value of human-machine teaming. "Differences between machine-generated recommendations and operator decisions highlighted the complementary strengths of human judgment, such as intuition, experience and situational awareness, in time-sensitive targeting," the service said.
Feedback from the exercise, as well as observations of how the AI operated, was then used to refine systems and processes. Former Secretary of the Air Force Frank Kendall said earlier this year that future warfare will be "highly automated, highly autonomous, action at long range, precision," and space will be a "decisive theater," adding that the "response times to bring effects to bear are very short"
"We're going to be in a world," he said, "where decisions will not be made at human speed; they're going to be made at machine speed."
And last year, he argued that mastery of artificial intelligence solutions could be key to winning the next major war.
US military branches have been testing how to use AI to speed up decision-making. The Air Force has been looking at pairing the technology with uncrewed systems and drones, like collaborative combat aircraft. The service has let AI algorithms fly fighter jets, and there has been a lot of time and research invested in how artificial intelligence can augment pilot capabilities.
AI as a tool can also complete back-end work in the military, such as better search functions for analyzing Department of Defense doctrine, understanding elements of specific locations, commands, or job positions, paperwork, and situation reports.
The rise of AI in militaries has been met with skepticism and ethical concerns from experts and officials about its implementation, especially in combat scenarios. The Pentagon has maintained that its policy on AI will keep a human in the loop for decision-making, though some observers have argued that this may not be possible in a high-speed, data-driven future war.
Others have noted that the technology may develop at a quicker pace than Washington or the Pentagon can regulate it. In recent Air Force testing, the human was kept in the loop. The AI functioned in a support role rather than hunting down targets independently with autonomy.
