Unexpected Findings in AI v Human Activity comparisons – Analysis
#AI #ArtificialIntelligence #MachineLearning #Defence #Technology
Recent findings by the US Defense Intelligence Agency (DIA) has revealed that in a real comparison
scenario where humans and AI were looking at enemy activity, Artificial Intelligence (AI) was actually
more cautious than humans about its conclusions of possible outcomes, especially when data is limited, turning conventional thinking on its head.
Artificial intelligence and humans have different risk tolerances when data is scarce.
While the recent findings are being classified as “preliminary”, they offer an important snapshot into
how humans and AI will complement one another in critical defence and security scenarios.
The DIAs technical director of Machine-Assisted Analytic Rapid-Repository System, [MARS] Terry Busch, revealed the experiments findings to Defenseone.com The initial phase examined the use of available data to determine whether a particular ship was in U.S. waters.
“Four analysts came up with four methodologies; and the machine came up with two different methodologies and that was cool. They all agreed that this particular ship was in the United States.” The conclusion is that Humans and machines using available data can reach similar conclusions.
The second phase tested a different methodology examining conviction. The question was, would humans and machines be equally certain in their conclusions if less data were available? For this phase, the test severed the connection to the Automatic Identification System, [AIS] which tracks ships worldwide.
Previous thinking had been that with less data availability, human analysis would be less certain in
its conclusions but the DIA Research has found the complete opposite. “Once we began to take away sources, everyone was left with the same source material — which was numerous reports, generally social media, open source kinds of things, or references to the ship being in the United States — so everyone had access to the same data.
The difference was that AI and those responsible for the machine learning, took far less risk — in
confidence — than the human analysis did. “The machine actually does a better job of lowering its
confidence than the humans do…”
The experiment provides a snapshot of how humans and AI will team for important analytical tasks, but it also highlights how human judgement can be clouded when pride is involved.