Fiscal Service Pilots AI Algorithm for Streamlining Appropriations Process
Without an AI tool, the bureau has to manually analyze the text of Congress' appropriation bills to determine which agencies get what amount of funding, Federal News Network reported Tuesday.
Fiscal Service Chief Data Data Officer Justin Marsico said the algorithm relies on natural language processing to read and interpret PDFs.
“There’s not an easy way that you can train a robot to pull out the right numbers. You need to understand syntax and the structures of the sentences and actual in order to pull the text apart," Marsico said during an Advanced Technology Academic Research Center summit.
Natural language processing is a branch of artificial intelligence aimed at understanding human language by drawing on principles from computer science and computational linguistics.
The bureau's AI algorithm turns document text into machine-readable data. Subject matter experts at the bureau then review the data to find and correct mistakes.
Marsico said the AI tool addresses the friction among employees caused by their teleworking condition during the coronavirus pandemic, particularly for office tasks such as physically signing documents.
Months into the pandemic, the Fiscal Service launched two projects aimed at streamlining the government's financial processes.
The first, called Digital End-to-End Efficiency, is meant to digitize and streamline the entire business process instead of applying individual technologies.
The second project, Blockchain for Grant Payments, is focused on using blockchain to let grant recipients tokenize, transfer and redeem payments.
Category: Federal Civilian
Tags: AI appropriations artificial intelligence automation blockchain Bureau of the Fiscal Service chief data officer Department of the Treasury federal civilian Federal News Network finance Justin Marsico natural language processing pilot