The U.S. Navy awarded Charles River Analytics a Small Business Technology Transfer (STTR) Phase II contract for the integration of human-AI teaming approaches into Navy operations.
Joint User-centered Planning Artificial Intelligence Tools for Effective Mission Reasoning (JUPITER), funded by the Naval Air Systems Command (NAVAIR), is an AI-based mission planning framework focused on the effective use of AI across Navy workflows and operational contexts.
Charles River Analytics explains that AI holds great promise, but must be thoughtfully integrated into operational workflows to reach its full potential. Simply applying off-the-shelf AI solutions to mission-critical problems can be ineffective and even counterproductive.
Integrating AI into workflows presents several challenges. End users need to understand when and how to distribute tasks across the human‑AI team, how AI reaches its decisions, and what is needed for effective coordination and execution.
While developing JUPITER, the team also focused on identifying potential operational scenarios for the Navy that are suitable for incorporating AI technologies, deriving requirements for human‑AI teaming within these scenarios, and developing methods to evaluate different AI solutions that might work within a human‑AI team.
Phase I of JUPITER involved understanding where and how AI would most significantly impact the mission planning process. The team established requirements for effective human‑AI collaboration in targeted operations and created lightweight simulation environments to evaluate concepts and simulate human‑AI interactions in realistic settings.
In Phase II, the JUPITER team built upon the prior success by identifying additional mission operations where human‑AI teaming could provide benefit to the Navy, and worked to develop, test, and validate these approaches. JUPITER is currently expanding in Phase II, focusing on transitioning the technology to target programs.
The key principles and innovative approaches developed under JUPITER for effective human‑AI teaming can translate to a number of commercial applications, especially as the interest in AI technologies (including generative AI) has continued to grow.
Stephanie Kane, Principal Scientist at Charles River Analytics and Principal Investigator on JUPITER, commented, “We need to take into account the human dimension when applying AI to critical problem spaces. For example, we must consider the capabilities and limitations of the AI technology, what the risks are, how it changes current workflows, where we should and should not use AI, and how the human and AI elements are going to effectively work together to accomplish overall goals.
“JUPITER is not just about identifying tasks that can be offloaded to AI. It also optimizes human‑AI interactions. The JUPITER framework considers the overall human‑AI team to cultivate effective interactions across the human end user and AI technologies. “It’s about evaluating their performance as a team and how the human’s work changes with the introduction of AI into the envisioned mission workflow.
“We’re deeply looking at the effective use of AI, especially considering the ultimate end users. The commercialization potential of JUPITER is immense, as human‑AI interaction is a fundamental challenge that numerous companies must address to leverage emerging and cutting-edge AI technologies effectively.”
