| Attachment | Size |
|---|---|
| AI Understanding.pdf | 3.1 MB |
Abstract
ACT-IAC White Paper: AI Understanding in US Federal Government
Developed by the Emerging Technology Community of Interest
Published February 15, 2023
In response to Executive Order 13960 “Promoting the Use of Trustworthy Artificial Intelligence in the Federal Government” (President, 2020), which requires agencies to implement AI Understanding, this paper provides an overview of Artificial Intelligence (AI) Understanding definitions, roles, taxonomy, modeling approaches, limitations, human-centered design enhancements, and a tailorable framework for planning and implementation. While this paper is intended for senior management, there is a wealth of additional information informing product managers, architects, machine learning engineers, product owners, data scientists, user Interface (UI)/user experience (UX) facilitators, and domain subject matter experts to accomplish the goal from a National Institute of Standards and Technology (NIST) paper on this topic:
“To build trust in an AI, those who evaluate the output of an AI for advice, decision making, auditing, validation, and related activities need to understand how the AI arrived at a particular answer. At a minimum, humans need to acquire a gist, a simple, yet productive, output explanation that captures essential or meaningful distinctions that human users need to make informed, insightful decisions.” (Broniatowski, 2021).