NTIS chief data scientist: Public-private partnership authority can help agencies with explainable AI

The National Technical Information Service helps agencies to improve data collection, labeling and classifying, and use data to ensure machine-learning models can be trusted.
United States Department of Commerce Building (Photo by James Leynse/Corbis via Getty Images)

Agencies looking to make better use of data for explainable artificial intelligence should take advantage of the National Technical Information Service’s “unique” partnership authority, said its chief data scientist.

Speaking during an ATARC webinar Thursday, Chakib Chraibi said Congress had the foresight to allow the National Technical Information Service (NTIS) to partner with top tech companies, academic institutions and nonprofits outside of the Federal Acquisition Regulation to address national data challenges and accelerate AI-based capabilities within all agencies.

NTIS is part of the Department of Commerce and works with agencies to determine how they are collecting, labeling and classifying, and using data to improve those processes and ensure machine-learning models can be trusted.

The agency stood up a Data Skills Working Group a few years ago and mapped skills for roles like data engineer, data scientist and data analyst.


The government wants the U.S. to be the global leader in what Chraibi said is the “premier technology of the 21st century,” going so far as to issue an AI Bill of Rights on Tuesday affirming its commitment to democratic values from development to deployment. But the field is evolving rapidly — machine-learning models becoming more explainable within weeks, not years — meaning agencies must improve their access to quality data.

“The issue is that a large number of federal agencies are still struggling with old data architectures that serve hundreds of applications in a vertically oriented, siloed approach,” Chraibi said.

Agencies also need mitigation policies in place beginning with prototyping to address AI risks like bias as they arise, Chraibi said.

A number of AI frameworks exist promoting responsible AI that agencies can choose from, but they can’t be implemented without metrics quantifying the explainability of the model.

“At NTIS we have a very agile framework that we work with,” Chraibi said. “And we work very tightly with the agency because they are the experts.”


Another area where agencies need to improve is ensuring they have the requisite data engineering and architecture skills on staff to modernize their infrastructure. Machine learning skills, while important, come later, Chraibi said.

“We try to know what are the skills that are within the Department of Commerce and what [employees] need to become a data analyst to upscale from within, as well as identify what we need from outside,” Chraibi said.

Latest Podcasts