WASHINGTON — The Pentagon’s clearinghouse on artificial intelligence is set to deliver a new tool to special operators to conduct psychological operations in near-real time in the information environment.
The tool, called Entropy, is meant to reduce the cognitive burden on personnel performing military information support operations, an official said during a virtual presentation on Sept. 10 during the Department of Defense AI Symposium.
In its simplest form, Entropy, under development by the Joint AI Center, will ingest data streams consisting of both text and video from what the military calls the information environment, or more commonly the internet, and provide the operators with summaries of trends based on the information.
This is what officials described as the passive portion. An active portion, which is still under consideration, would build upon the passive component by taking the topics identified, feeding them into a language model, and downselecting messages for the team to review and score. This would help train the algorithm to improve. This version would also push those messages into the information environment.
Given the pace of information in today’s environment, operators need to be able to ingest content and rapidly respond with counter messages or shaping messages of their own. Entropy seeks to facilitate forces to compete in this dynamic environment in near-real time.
“As we look at the ability to influence and shape in this environment, we’re going to have to have artificial intelligence and machine-learning tools, specifically for information ops that hit a very broad portfolio,” Gen. Richard Clarke, the head of of Special Operations Command, said at the Special Operations Forces Industry Conference in May.
“We’re going to have to understand how the adversary is thinking, how the population is thinking, and work in these spaces in time of relevance. If you’re not at speed, you won’t be relevant," Clarke added. "What we need is adapting data tech that will actually work in this space and we can use it for our organization.”
A minimum viable product for Entropy has already been built and is undergoing testing. The system was designed for Special Operations Command Pacific and focuses on three languages: Tagalog, Mandarin and English. Officials said it could be deployed to another region, but it would require language retraining.
The architecture also allows for third-party integration into commercial or government-built platforms, according to the presentation at the AI Syposium.
The Entropy project is part of the Joint AI Center’s newest pillar, dubbed Joint Information Warfare. Previously, this was the national cyber mission initiative, but has now expanded to include information operations and electronic warfare.
It’s unclear what electronic warfare efforts have taken place under Joint Information Warfare, but previously artificial intelligence and machine-learning initiatives in the electronic warfare space across the DoD have zeroed in on faster signal classification and improving the department’s understanding of the electromagnetic spectrum for operational planning.
Joint Information Warfare’s goals are to provide “AI-enabled information solutions to DoD and the joint force that will contribute to information advantage in all domains and the EMS through assurance, resilience, flexibility and security,” according to the presentation.
“Externally, our biggest challenge is the scope of our mission space; cyber, IO and EW each represent large joint disciplines. Clearly understanding the needs of each community and, more importantly, identifying where AI can have a mission impact is critical to our success,” said Col. Pete Tortorici, director of the new pillar.
Mark Pomerleau is a reporter for C4ISRNET, covering information warfare and cyberspace.