Around one-fifth of federal agencies use artificial intelligence to help process public records information requests, the National Archives and Records Administration (NARA) revealed in a new report.
NARA officials warned that while AI can help with Freedom of Information Act (FOIA) requests, the technology shouldn’t substitute human decision-making in determining exemptions due to possible harm in those processes.
New research from earlier this year found that nearly 19% of agencies use AI or machine learning (ML) in processing FOIA requests, and that around 78% of agencies said they don’t use those technologies to help with searching public records.
“AI and machine learning are revolutionizing the way information is stored and processed across all sectors of society,” said NARA officials.
While most agencies haven’t begun to use AI/ML yet, NARA said that “the early adopters are demonstrating the ability of AI to identify sensitive information and normalizing the concept of AI in FOIA processing.”
As more agencies adopt AI to help with their FOIA processes, NARA said the use of the technology should maintain human oversight.
“AI and machine learning have the potential to aid in FOIA processing but are not a substitute for the judgment of FOIA professionals on application of exemptions and foreseeable harm,” said NARA officials.
The report added that while agencies should use caution when using AI, it is also “important that agencies explore the use of AI and/or machine learning options to help improve FOIA processing response times.”
Meanwhile, nearly 74% of agencies said they use e-discovery tools – software used to find, review, and produce electronically stored information – when responding to FOIA or legal discovery requests, a number that hasn’t changed much in recent years.