Biden administration working on ‘enhancing’ AI use case reporting, Martorana says

The Biden administration’s efforts to improve reporting of artificial intelligence use case inventories include efforts to make them more searchable, the government’s top IT official said Tuesday.

“We’re working really hard to make sure that we’re enhancing those use cases … with metadata so that we can search them and really interrogate them, rather than just collect them and broadcast them — really to get key learnings from those,” Federal CIO Clare Martorana told reporters at a Federal CIO Council symposium Tuesday. 

The White House has previously indicated the inventories will be more central to understanding how agencies are using the technology going forward. In fact, draft Office of Management and Budget guidance that corresponded to President Joe Biden’s AI executive order proposed expanding the inventories with information about safety- and rights-impacting AI, the risks of uses, and how those risks are being managed.

“Federal agencies have special responsibility to get AI governance right, and we believe this policy will continue our global leadership,” Martorana said of that guidance in a keynote address earlier in the day.

The draft guidance was released in November shortly after Biden’s AI order and would establish a framework for agencies to carry out the administration’s policies for the budding technology. It included, among other things, requirements for agencies to designate chief AI officers — which agencies have already been starting on — and expanding existing reporting on agency AI uses.

Martorana, while talking to reporters, said public comment was “critical” to the development process for the guidance, and noted that equity and transparency as common themes in comments they received from interested parties.

With respect to transparency, Martorana pointed to the administration’s desire to improve agencies’ AI use case inventories, which were required initially under a Trump-era executive order and later enshrined into statute.

As of September, the federal government reported over 700 public uses of AI, demonstrating broad interest and potential for the technology across the federal government. Those inventories, which are required annually, have also so far been inconsistent in terms of things like format and information included.