Government officials at both the Federal and state levels are working on a wide range of new AI use cases with focuses on administrative efficiency and improving decision-making, several Federal and city officials said this week.
Speaking at the ATARC 2024 Public Sector Summit on Dec. 19, officials from the Social Security Administration (SSA), National Science Foundation (NSF), National Oceanic and Atmospheric Administration (NOAA), Office of Personnel Management (OPM), and the City of Boston shared their latest AI use-case initiatives.
Santiago Garces, the chief information officer (CIO) of the City of Boston, said that recent pushes in AI include using generative AI (GenAI) and computer vision to create digital representations of parking restrictions to improve policy planning and provide better customer experiences. The city also is looking to use a GenAI system to help with administrative tasks such as creating simplified summaries of council meetings.
Other prototypes being developed by the city aim to enhance searches on the Boston.gov website by leveraging retrieval-augmented generation and vectorized data to provide more accurate answers. Additionally, the city is looking to utilize Google Workspace’s AI Gemini to boost productivity among city employees.
“Generative AI as a general-purpose technology has the possibility of improving the productivity of all 25,000 employees of the city … [but] we have very different people that do very different kinds of jobs … and that they might need slightly different versions of software,” said Garces. “We’re working on a very experimental approach … to see if different people take up the version of [Google’s AI model] Gemini … [and] what is the impact.”
NOAA has been working its AI use cases for a wide range of scientific inquiries – from “the surface of the sun to the bottom of the ocean,” shared Frank Indiviglio, the agency’s chief technology officer.
“AI has a potential of really allowing greater resolution, higher high-resolution models, which gives you better data, which gives everybody a better forecast, but there are really kind of interesting use cases within that sphere,” said Indiviglio about using AI to improve environmental monitoring and management.
“These kinds of methods can be applied across the Earth system – modeling, observing – and they’re really exciting things, and they can really help people do their job better … A lot of the technologies that we’re looking at … allows them to do more in the field while they’re there,” he said, adding, “We’re expanding the workforce without expanding the workforce.”
NSF – which is approaching its 75-year anniversary next year – is applying AI models to sift through decades of collected data to understand the impact of billions in research funding and to make improved investment decisions in the future, Michael Hauck, the responsible AI official and deputy chief data officer at NSF, explained.
“How do we take the 75 years of data that we have and tease out from that what we as taxpayers got for our investment – that’s key on our minds,” said Hauck. “AI is a tool that’s going to help us do that, because all that 75 years of data is in different formats, collected for different reasons … So how can we tease out from all of that, what we got for our money, and when we could do that, we understand that, then that can help us make better investments going forward.”
AI can also be used to convert legacy code – such as code written in COBOL – to more modern languages such as Java or Python, explained Melvin Brown III, the deputy CIO at OPM. Brown said that the office has been working to convert its existing mainframe to modern languages that can be maintained in the long term.
“We’re excited about where we think that that work will go, and then starting to reimagine just some of our back-office support services – from our help desk to our finance or HR – and how can we use some of the built-in AI capabilities to get more productivity out of employees that we service,” he said.
Similarly to OPM, SSA has been working on AI use cases to improve administrative and back-office functions, SSA CTO Sid Sinha said. Some AI use cases aim to automate paper-intensive processes such as disability claims processing and routing – with some undisclosed AI systems already in use, Sinha added.
“How can we speed [data processing] up for medical information and correlation with unique cases and claims that cases have,” said Sinha. “There are lots of opportunities to really make those much more seamless.”
One of the agency’s newest pilots also includes a ChatGPT-like chat bot for internal use, following a new trend of internal AI chatbots after the Department of Homeland Security announced its DHSChat bot earlier this week.
“We are working on our own sort of private Chat GPT … for different sort of scopes or bodies of knowledge, and we’ll be doing several of those in this coming year, we’re working on them right now” Sinha said.
