U.S. Army Capt. Michael Holder, a military police officer assigned to V Corps, analyzes a computer during Allied Spirit at U.S. Army Garrison Bavaria Grafenwöhr on March 6, 2023. (U.S. Army photo by Pfc. Myenn LaMotta)

GVSETS 2024 — The US Army is upping its bet on artificial intelligence and working to broaden acquisition plans in hopes the cutting-edge, at-times-controversial tech can help with everything from sensor operations to ground vehicles to aviation and training, according to one of the service’s top tech officials.

“We are doing things now so we can accelerate that adoption,” Young Bang, Principal Deputy Assistant Secretary of the Army, Acquisition, Logistics, & Technology (ASA(ALT)), said Wednesday during a panel at the 16th Annual Ground Vehicle Systems Engineering & Technology Symposium (GVSETS) in Novi, Mich. “We want the ability to make risk-based decisions with some controls.”

Meanwhile, the Army is also looking to AI defenses as well, according to Jennifer Swanson, SES, Deputy Assistant Secretary of the Army (Data, Engineering, & Software). She told Breaking Defense, “We have to protect ourselves from whatever our adversaries will bring to the table.”

Under the Army’s ongoing #DefendAI initiative, the service is seeking support from companies and academia, including a July Request for Information (RFI) to better understand the industry’s capabilities and gather information and related to what it’s calling AI-LDF for AI and layered defense.

The branch is also interested in risks associated with traditional adversarial methods, such as data poisoning and model stealing, in addition to emerging and future menaces with the potential for security disruption. This data will be used to inform requirements for competitions and awards in the coming years.

RELATED: ‘Poisoned’ data could wreck Ais in wartime, warns Army software acquisition chief

Swanson stressed the goal is ensuring that the Army has “the big picture” and that the service is “not missing pieces or not asking industry to do things that aren’t achievable”.

“By the end of August, we should have all those responses in. It will take us a little bit of time to go through it and analyze it, but the idea is to evolve what we put out based on the feedback that we get,” she added. “I don’t know exactly what we’re going to get”.

Another Army initiative that could benefit from the approach is Project Linchpin, in which the branch intends to integrate AI and machine learning (ML) into weapons and sensor modernization programs while protecting operational data.

Announced in 2022, Linchpin is the Army’s first program of record that is expected to deliver AI and ML capabilities at scale. It is intended to enable faster processing, exploiting, and dissemination of intelligence data, according to the Program Executive Office Intelligence, Electronic Warfare & Sensors (PEO IEW&S).

In October 2023, the branch awarded Linchpin contracts to Booz Allen Hamilton and Red Hat to support research efforts in terms of Traceability, Observability, Replaceability, and Consumability (TORC) to guarantee model and data integrity, data openness and modular open system architecture (MOSA) design.

Swanson highlighted that other Linchpin agreements should be announced in the next couple of years. “There is potential for other programs to put out contracts [as well]. We are working through that as part of our 500-day plan,” she said.

In March 2024, the ASA(ALT) announced its AI Implementation Plan which involves multiple efforts within 100- and 500-day execution window. Planned to establish an AI foundation for future programs, the 100-day portion was recently concluded. Meanwhile, the ongoing 500-day initiative is focused on identifying capabilities gaps across the industry.

“There has certainly been a huge amount of progress in the commercial side for the past couple of years, but there are still a lot of things left to be done,” Swanson remarked.