Posts

From NTDS to JADC2: The Unsolved Problems of Multi-Sensor Fusion

Image
What AI Cannot Fix Proceedings U.S. Naval Institute  |  Analysis & Commentary C4ISR & Sensor Fusion The Naval Tactical Data System's most intractable operational problems — sensor alignment bias, dynamic navigation error, and multi-sensor track association failure — were substantially ameliorated by the E-2C/D Hawkeye's network-derived registration and the Cooperative Engagement Capability's architectural innovation of distributing raw sensor measurements rather than processed tracks. Neither fully solved the problem, and neither transfers completely to the GPS-denied, multi-domain, multi-service environment that JADC2 requires. The AI system at the top of the processing chain receives a track picture corrupted by errors it cannot see, and generates confident engagement recommendations it cannot qualify. Stephen L. Pendergast, LT USNR, IEEE Sr. Member (Ret.)  |  Senior Engineer Scientist J...

Hallucinating the Firing Solution:

Image
  Large Language Models in Autonomous Weapons and the Coming Fratricide Incident Proceedings U.S. Naval Institute  |  Analysis & Commentary Vol. 152  |  June 2026 Technology & Weapons Systems Autonomous Systems & Artificial Intelligence The Department of Defense is integrating Large Language Models into autonomous weapons architectures faster than testing methodology exists to validate them. The specific failure modes of LLMs — hallucination, non-determinism, helpfulness-driven completion of missing data, and prompt injection vulnerability — are not engineering defects that will be resolved through iteration. They are properties of the architecture. A fratricide incident is not a risk to be managed. It is a timeline to be shortened or lengthened by the decisions made in the next eighteen months. Stephen L. Pendergast, IEEE Sr. Member (Ret.)  |  Senior ...