The US Defense Advanced Research Projects Agency (DARPA) has launched a program to explore the use of algorithms in helping to make critical military decisions under challenging circumstances.
Called the In the Moment (ITM) program, the initiative would quantify the alignment of algorithms with authorized human decision-makers to help when there are no agreed-upon answers.
The algorithms can assume decision-making when human decision-makers disagree or during uncertainty, resource limitation, or time constraints.
“Two seasoned military leaders facing the same scenario on the battlefield, for example, may make different tactical decisions when faced with difficult options,” DARPA said in a press release. “As AI (artificial intelligence) systems become more advanced in teaming with humans, building appropriate human trust in the AI’s abilities to make sound decisions is vital.”
The agency further stated that understanding the key characteristics of expert human decision-making and representing them in algorithmic decision-makers is essential to ensure algorithms make trustworthy choices.
The ITM program includes the development of decision-maker characterization techniques and the creation of a quantitative alignment score between a human decision-maker and an algorithm.
“We’re going to collect the decisions, the responses from each of those decision-makers, and present those in a blinded fashion to multiple triage professionals,” ITM program manager Matt Turek said. “Those triage professionals won’t know whether the response comes from an aligned algorithm or a baseline algorithm or from a human.”