The National Science Foundation (NSF) is experimenting with AI to help its employees and closing in on the release of a new chatbot for the agency’s outside constituencies, NSF Chief Data Officer and AI Official (CAIO) Dorothy Aronson said this week.
As part of a broader IT reorganization, NSF announced early this year the establishment of an independent Office of the Chief Information Officer (OCIO), which houses its new Division of Data and AI (DDAI).
“The change in the organization really came because NSF is growing,” Aronson explained during a NAPA webinar on April 22. “We had a medium-size IT shop, and it needs to be a lot bigger.”
“The more work that’s done with computers – through AI and everything and supercomputing and whatnot – we need more people who are educated in those fields,” she said. “We grew by appointing a new CIO and a segment of his organization is now devoted to just data and AI, which wasn’t true several years ago. So, we have a whole new class of workers that need to be included in our environment.”
The CAIO explained that her focus at NSF is on AI within the operational space.
“The AI that I’m most focused on is providing AI for the workforce of NSF to improve the business processes and to help customers who are trying to access NSF get to what they need as quickly as possible and efficiently,” she said.
Aronson said her team is currently “very rapidly” developing an AI chatbot that will be externally facing to help the community understand how they can best work with NSF.
“We asked NSF, ‘What use cases can you think of that we could implement?’ So right from the start we engaged the community in the discussion,” she said. “Within a couple of weeks, we got about 80 ideas.”
“The thing we’re building is a chatbot that will be externally facing to the community and will answer questions about, ‘Well how do I work with NSF?’ or ‘I’m a physicist, which solicitation or which program would be a good one for me to connect with?’ It’s multilingual, and so it’s going to help people answer these simple questions – especially people who don’t have large organizations behind them that are prepping them and preparing them for work with NSF,” the CAIO said.
Aronson said the chatbot is a six-month development effort.
“We’re dead in the center right now,” she continued, adding, “In a few months, by the end of the spring, we’ll have something in hand.”
