Washington, D.C. – December 12, 2023
In a comprehensive report released by the Government Accountability Office (GAO) on Tuesday, it was revealed that the United States government’s plans to fully embrace artificial intelligence (AI) lack a cohesive strategy. The report, spanning April 2022 to December 2023, delves into the current state of AI adoption across federal agencies, shedding light on successes and challenges.
The GAO report underscores the uneven progress among government bodies regarding AI integration. While agencies such as NASA and the Department of Commerce have made substantial investments in AI, others are grappling to meet federal requirements, exposing a lack of a unified approach.
The GAO’s extensive survey of 23 non-defense agencies unveiled a total of 1,241 current and planned AI use cases across 20 different agencies. Notably, agencies like NASA and the Department of Commerce emerged as leaders, boasting 390 and 285 AI use cases, respectively. However, the report also pointed out that only about 200 of the identified use cases were currently operational, while 516 fell into the planned category.
The documented use cases highlight the diverse applications of AI technology within the federal government. NASA, for instance, utilizes AI for intelligent targeting and selection of scientific specimens collected by planetary rovers. Meanwhile, the Department of Commerce employs AI to automate the counting of seabirds from drone photos.
A significant challenge identified in the report is the lack of a standardized definition for AI systems or technology. The report notes that both the scientific community and industry lack a common definition, leading to varying interpretations even within the government. Regulatory enforcement, the report suggests, may face difficulties due to disparate definitions found in federal statutes.
Furthermore, the GAO discovered issues with the data submitted by agencies to the “Office of Management and Budget (OMB).” Out of the 23 agencies surveyed, only five provided comprehensive information on their reported AI use cases. Fifteen agencies submitted incomplete or inaccurate data, ranging from incomplete fields to duplications. Notably, two agencies had initially labeled non-AI inventories as AI.
While some agencies, including the Department of Commerce and General Services Administration, were recognized for fully implementing selected AI requirements, the report highlighted that 12 agencies had only partially met the specified criteria. For example, the “White House Office of Science and Technology Policy” had not communicated expectations regarding AI requirements for specific agencies.
The report concludes with recommendations for the “White House Office of Science and Technology Policy (OSTP), Office of Management and Budget (OMB), and Office of Personnel Management (OPM)” to implement AI requirements with implications for the entire government. Additionally, it advises the 12 agencies to fully implement specific requirements, while urging the 15 agencies with incomplete AI inventories to prioritize data updates.
While the majority of agencies either agreed or partially agreed with the recommendations, OSTP and OMB disagreed on some points. They argued that the White House’s AI executive order, released after the GAO’s draft report, superseded previous guidance, impacting the requirements placed on them. The report thus highlights the evolving landscape of AI governance within the federal government.