Florida Air National Guard, 125th Medical Group Detachment 1, set up patient treatment tents, practiced triage, and ran search and rescue operations at Camp Blanding, Florida, September 20, 2020. (U.S. Air Force photo by Senior Airman Cole Benjamin, Released)
By Cal Biesecker |
@calvinb21
10 months ago |
06/06/2023
highlights

The Defense Advanced Research Projects Agency (DARPA) on Tuesday announced five performers to develop a framework for trustworthy artificial intelligence (AI)-based algorithms to make decisions in moments when every second counts and there is no right answer and decision-makers disagree.

The selectees will conduct their work in two phases, the first being small unit triage in austere environments and the second phase being mass casualty triage.

The “In The Moment” (ITM) program performers include:

  • Raytheon Technologies’ [RTX] BBN Technologies unit and Soar Technology, Inc., which will each develop decision-maker characterization techniques to identify and quantify key human decision-maker attributes in difficult domains. DARPA awarded Raytheon $3.8 million and Soar $4 million;
  • Kitware, Inc. ($6.2 million) and Parallax Advanced Research Corp. ($4.1 million) will each develop algorithmic decision-makers that demonstrate alignment with key attributes of trusted human decision-makers; and
  • CACI International [CACI] ($6.6 million) will design and execute the program evaluation, focusing on how key human attributes may lead to the trusted delegation of decision-making.

The Univ. of Maryland Applied Research Laboratory for Intelligence and Security and the Institute for Defense Analyses are supporting the ITM effort under an existing task order and will focus on policy and practice integration and outreach, including ethical, legal, and societal implications experts to advise throughout the research process.

“Key attributes might include how an algorithm evaluates a situation, how it relies on domain knowledge, how it responds to time pressures, and what principles or values it uses to prioritize care,” Matt Turek, DARPA’s ITM program manager, said in a statement. “From a technical standpoint, difficult decisions made in medical triage will likely require approaches that do no rely primarily on training data for their implementation, as those approaches can be notoriously brittle.”

Turek said that producing a framework around the triage domain gets right to “core issues around trust and delegating decision-making to go beyond the state of the art in AI.”