Artificial Intelligence and Machine Learning Basic Research

The Artificial Intelligence and Machine Learning Basic Research program is a foundational Army research initiative focused on advancing artificial intelligence (AI) and machine learning (ML) technologies to support an AI-enabled Multi-Domain Operations (MDO) Force. The program executes both intramural and extramural basic research, targeting breakthrough developments that can impact critical Army missions such as target detection, autonomous maneuver, predictive maintenance, intelligence support for operations, network/cybersecurity, business process automation, and medical support. Strategic guidance and coordination are provided by the Army's Artificial Intelligence Integration Center (AI2C), ensuring alignment with the Army Modernization Strategy and the Chief Digital and Artificial Intelligence Office.

The AI/ML Basic Research Hub is a central component of this program, operating as a consortium that brings together academia, industry, and government. The Hub, located at Carnegie Mellon University, leverages the strengths of each sector academic innovation, industrial transition expertise, and Army-focused problem-solving to accelerate the development and fielding of AI/ML capabilities. Its research portfolio addresses Army-relevant areas such as object recognition using Multiple Cooperative Autonomous Sensors (MCAS), leader decision-making, autonomous tactical behaviors, robotics, network resiliency, and medical support. Collaboration within the Hub is designed to ensure that research outcomes are both innovative and directly applicable to Army needs.

Within the Hub, the Intelligence Support to Operations line item focuses on developing AI/ML methodologies for object detection on imagery to augment operational capabilities. This research addresses the challenge of recognizing surrogate targets in test ranges, using AI trained on real operational objects, and supports long-range precision fires. The goal is to enhance the Army's ability to identify threats and provide actionable intelligence in complex operational environments.

The Foundation Models line item centers on advancing machine learning models that can generalize across tasks and domains. These models, which include generative methods, are trained on vast datasets and are designed to enable rapid development of accurate solutions for a variety of Army applications. Research in this area seeks to improve techniques for transferring foundation models to new domains, synthesizing multi-modal data, and enabling use-cases such as natural language querying, semantic segmentation, and automated threat recognition. The objective is to unlock robust, adaptable AI tools that can address current and emerging Army challenges.

The Distributed AI line item aims to improve the Army's ability to train, deploy, and govern AI/ML models across both enterprise and tactical environments. This includes research into federated learning, deploying algorithms on ruggedized edge hardware, and enhancing robotic autonomous systems. Security is a key focus, with efforts to develop techniques for both attacking and defending AI/ML systems. The goal is to ensure that distributed AI architectures can operate securely and efficiently, even in contested environments, and that they can autonomously adapt to dynamic resource availability.

The Human AI Interactions line item addresses the integration of AI/ML systems with human decision-makers, particularly in high-stakes, complex environments. Research in this area explores how to make AI outputs more interpretable, evaluate the safety of human-AI interactions, and ethically apply AI to decision-making processes. Additional objectives include developing effective training for users with diverse technical backgrounds and leveraging AI to process and summarize large datasets for human consumption. The aim is to enhance the Army's ability to deploy and utilize AI/ML products in operational settings.