Federal Agencies Learn to Embrace AI

November 09, 2020

Among the top-line statistics in the “Artificial Intelligence in Federal Administrative Agencies,” report that was developed by researchers at Stanford University and NYU, is that 45% (64) of the agencies canvassed indicated interest in AI or ML through their activities. 

By way of example, says the report, high-impact weather tracking systems are being used by the National Oceanic and Atmospheric Administration (NOAA)  “to improve decision-making in real time.”

We also learn that AI-based tools to predict health care fraud are being developed by the Centers for Medicare and Medicaid Services. 

The seriousness of purpose by the federal government to institutionalize the practice and the advancement of data governance within its ranks is codified in the Foundations for Evidence-Based Policymaking Act of 2018.

Designated H.R. 4174, it was signed into law by U.S. President Donald Trump on Jan. 14, 2019. The law requires federal agencies to make data accessible and machine-readable, to develop evidence-based policy and evaluation plans, and to designate Evaluation Officers, Statistical Officials, and Chief Data Officers to support and implement the new requirements. 

H.R. 4174 also incorporates the Open, Public, Electronic, and Necessary (OPEN) Data Act. 

The head of each agency is required to designate an employee as the agency’s Chief Data Officer (CDO). The position is responsible for the provisions articulated in the OPEN Government Data Act. The Chief Data Officer also serves as a member of the Chief Data Officer Council, operated under the aegis of the Office of Management and Budget (OMB).

The Council’s charter is to establish government-wide best practices for data creation, use, protections and sharing, and to promote data-sharing agreements between agencies.The Council is expected to submit a biennial report to the Director of OMB and to Congress.

Other roles of the CDO Council are to evaluate new technologies for how federal agencies gather and implement data, advocate standardization protocols for data management across the enterprise, and coordinate with other government councils on data-related activities.


The legislation (H.R. 4174) was lauded by the Center for Data Innovation’s Daniel Castro as “a major achievement for the open data community,” adding,“ by passing this law, the United States has established itself as one of the global leaders in open data.”

Agencies were mandated to appoint a CDO by mid-2019, but even into late 2020, results of that directive were still in flux. Some agencies already had a CDO position established years before, others added the role by hiring a new person, and still others took the so-called “dual-hatted” route, adding the CDO responsibility to an existing position, typically the Chief Information Officer (CIO) or Chief Financial Officer (CFO). 

In a January-February 2020 survey of CDOs co-sponsored by the Data Foundation, Grant Thornton, and Qlik — to which one in four CDOs responded — roughly one-third of CDOs said they report to a CIO. Others report to various officials, be it agency heads or Chief Operating Officers (COOs).

CDOs who report to a CIO tend to have smaller staffs than CDOs reporting to other positions. Forty-three percent have been in the CDO role under one year, 25% for one to two years, and 32% more than three years. More than half said their staff has fewer than 10 persons, 21% said they have 11-25 employees, and 25% have more than 25. 

Government workers appointed to the CDO position tend to have veteran status, with 60% employed in their current federal agency more than five years. 

While the vast majority (86%) said they understood what was expected of their position, as articulated in the Policymaking Act and the Federal Data Strategy 2020 Action Plan, only slightly more than half said they knew how to meet those goals. 

Limited budgets were cited by more than six in 10 CDOs as a problem in getting their work done satisfactorily, while fully half were not clear on the exact scope of their job description. 

There was encouraging news, as well, to draw from the survey. 

Nearly 80% of CDOs surveyed cited successful steps to build inventories of data assets. 

Three in four CDOs identified positive outcomes from their data governance activities. More than 60% saw an increase in data quality. Others benefited from assessing staff capabilities and needs, migration to cloud-based services, and availability of metadata.