Academics from the UK and US have piloted the application of artificial intelligence to aid combat healthcare judgments, investigating the readiness of individuals to rely on and assign critical decisions to AI.
This project, spearheaded by the Defence Science and Technology Laboratory in partnership with the US Defense Advanced Research Projects Agency, delved into methods for integrating AI with human judgment processes in demanding situations, as reported by the UK Defence Journal.
These experiments leveraged DARPA’s “In the Moment” initiative, which centers on adapting AI systems to embody individual human inclinations and main objectives. A key challenge in AI implementation is that these systems do not inherently reason or act like humans, and researchers are striving to resolve this by finding a recognized method to quantify or reproduce human judgment in intricate situations.
Mock large-scale injury incidents were employed for the experiments, which occurred in October 2025 at Merville Barracks in Colchester and RAF Brize Norton. Initially, participants underwent evaluation to ascertain their individual judgment patterns, including how they weighed considerations like preserving the maximum quantity of lives, emphasizing quality of life, or preferring particular individuals based on their association or function.
Subsequently, AI systems were programmed to either reflect or intentionally deviate from these inclinations, thereby functioning as a primary healthcare provider within the situation. Individuals participating examined the AI’s sorting judgments and were queried if they possessed sufficient confidence to assign responsibility. Their engagement with AI was not disclosed until the conclusion of the exercise.
The primary goal of these experiments is to investigate if harmonizing AI with human principles boosts confidence and the readiness to assign judgments. Should they prove successful, such systems might empower healthcare providers to manage and categorize greater volumes of injured individuals with greater speed, possibly elevating recovery probabilities in demanding operational settings.
“We’re looking at human-AI teaming in a medical triage setting,” stated a Dstl human factors expert who contributed to the project. “We’re really interested in how the warfighter makes decisions based on increasing amounts of information and how AI systems can support that.”
The examination of the findings is currently underway and will inform subsequent studies concerning human-AI collaboration and judgment processes within the defence sector, particularly regarding how such systems could be securely implemented in field environments.