New research fro the Defense Intelligence Agency (DIA) reveals that AI technology can be more cautious about its conclusions in situations where data is limited than humans are. In a situation in which humans and AI were assessing enemy activity, the AI technology was more cautious in the decision making process. Although the results of the study are still preliminary, they offer insight into how the future of AI and human collaboration in national security fields will look.
The study involved testing humans and AI in a situation where they must come to conclusions when given a very limited set of data. With less data, it is probable that the human analyst should be less certain in their conclusions than AI, however, the researchers found that the opposite occurred. When AI and humans were given a small amount of data and asked to reach a conclusion on a national security situation, the AI machines were warier of the lack of information than humans.