FormAID: an AI bridge between form authors and form receivers

In order to maintain its military advantage in a digitally competitive world, the US Navy must embrace AI technologies to stay ahead of evolving threats from near-peer adversaries. Operational efficiency is one area that could significantly benefit from the application of novel AI solutions. In particular, AI assistance in form completion documentation tasks could greatly increase operational efficiency. To help the Navy develop this potential, Pacific Science & Engineering Group, Inc. (PSE) proposes an AI-powered platform that employs cutting-edge natural language processing (NLP) technology to greatly enhance the speed and quality of electronic form submissions. Our proposed solution, FormAID, would consist of a language model-based AI platform that integrates form guidance policies and theory of mind (ToM) principles from cognitive science to assist form-fillers in generating entries that maximize form-receiver understanding. Scientifically uninformed solutions to the form completion workflow problem are likely to solely focus the form-author role with NLP and other tools for them. PSE, on the other hand, recognizes that a holistic user-centered design (UCD) process of the team processes is needed to fix a team problem. Thus, we propose to develop FormAID, which will involve gaining a detailed understanding of the needs of both the form-author and form-receiver. Our contention is that if the form-author does not have a good ToM for the form-receiver (i.e. they don’t understand the needs and context of the form receiver, including their training, expertise, knowledge of various materials, awareness of specific fleet memos, part advisories, recalls, familiarity with jargon, shorthand, and specialized terminology, etc.) they are likely to produce form entries that do not contain enough information or provide excessive irrelevant information triggering the inefficient, iterative workflows. Once the needs of both the author and receiver are firmly understood, we will engineer a language-model based platform that engages with the form-author in ways that mitigate the ToM problem, thereby ensuring form data is clear and understandable from the perspective of the form-receiver. Furthermore, our platform will validate form data to ensure it complies technical guidelines. Overall, the FormAID platform aims to enhance the form completion workflow, and by so doing, increase Navy material readiness. PSE's unique combination of expertise in human-AI teaming, human-system interface (HSI) design, and UCD principles positions us to deliver a practical and innovative solution that addresses and fixes the motivating workflow problem.