The State Department and the National Institutes of Health rely on artificial intelligence tools to accomplish their missions and gain more insights from their data in an increasingly interconnected world.
Matthew Glavis, the State Department's chief of data and AI, said the department views AI as: ““It is an essential tool in our diplomatic toolbox,” and we are taking steps to ensure guardrails to ensure the responsible use of this technology.
“There is a strong need to identify opportunities, take advantage of them where possible, and use them responsibly. The world we live in now is more interconnected than ever before. And that applies to diplomacy as well. Many of today's diplomatic challenges transcend national borders. They are global, like climate, for example. It is a global challenge,” Gravis said. “This is not specific to any particular country; our country engages with other countries bilaterally on certain matters. We see an opportunity for generative AI to create significant advantages in synthesizing large amounts of information, given the need to understand the local and global context.”
The State Department released its first-ever enterprise AI strategy in November 2023, shortly after the Biden administration announced a sweeping executive order to strengthen the use of AI tools in government.
“We took an all-department approach to developing our strategy. We engaged multiple departments within the State Department to participate in the process to ensure that factors such as privacy, cybersecurity, and legal considerations were considered. ,” Gravis said.
The AI strategy focuses on four goals: secure infrastructure, culture, and responsible AI adoption and innovation. Graviss said much of his culture efforts rely on employees sharing his AI success stories across the department.
“Analysis without communication, if you don't share what you're doing and how you're doing it, you're literally helping one person. You're just helping one person. ,” Gravis said. “It is critical that we build a culture that these tools, when used safely, can help us operate more efficiently as a government and improve our diplomatic outcomes.”
Mr. Graviss leads the department's AI Steering Committee, which focuses on AI policy and guardrails, as well as legal, privacy, and security.
NIH lowers barriers to AI research
Susan Greglick, NIH's associate director for data science and director of the Office of Data Science Strategy, leads data science and AI efforts across NIH's 27 institutes, centers, and offices.
“We are working to ensure that new innovations in trustworthy AI and approaches that reduce bias and risk fundamentally fund biomedical research, ensure data fairness, and improve model We're also working to deliver testable and explainable AI algorithms, which are very important areas that impact the biomedical and healthcare research enterprise,” Greglic said.
NIH is recruiting data scientists and AI computer scientists through the Data Scholars Program. The program will bring together experts from academia, industry, nonprofit organizations, and other federal agencies during his two-year sabbatical at NIH to lead cutting-edge research.
Through the AIM-AHEAD program, NIH also strives to advance health equity and promote diversity among researchers developing and using AI tools.
“When you think about AI and the NIH, think about trustworthy, fair, and unbiased AI that impacts the broadest possible community,” Greglick said.
Greglick said the administration's AI executive order calls on NIH to strengthen and expand the AIM-AHEAD program.
“For students, from middle school students to tenured professors, the AIM-AHEAD program offers an opportunity to gain hands-on experience with artificial intelligence and develop capabilities and capabilities in underrepresented and underserved institutions. There will be more,” Greglick said.
The NIH is working on “multimodal” AI trained on data ranging from medical records to medical images, wearable sensor data and genomics data, she added.
“Multimodal AI has the potential to improve predictions and provide new insights into highly complex biomedical systems. and multidimensional integrated systems,” said Greglick.
The NIH is pursuing an “ethics first” approach to multimodal AI in biomedicine, and is taking steps to ensure that AI tools produce ethical and unbiased results, she added. .
“As we think about new work in AI and new approaches to funding AI research, this is something we will push forward with,” she said.
NIH also supports federal efforts to make AI resources more accessible to the broader research community. The agency provides datasets to the National AI Research Resource (NAIRR), a hub for federal AI research resources.
The Biden administration says that by lowering barriers to participation in AI research and allowing more researchers to participate in AI development, the research community will be better equipped to tackle new challenges around the ethical use of AI. Are expected.
“We are contributing a significant number of high-quality datasets and imaging in immunology, cancer, multiple chronic diseases, and data from healthcare settings,” Greglick said.
NIH is partnering with the Department of Energy on the NAIRR Secure pilot to give researchers access to data from NIH medical and clinical imaging sites while keeping the data secure.
Generation AI as a “true force multiplier”
Gravis said one example of the State Department's use of generative AI is to help diplomats quickly understand local news produced in the countries and regions where they are stationed.
“We have a responsibility to understand what is happening in a particular post. If you work in Tashkent, or if you work in Tegucigalpa, we have a responsibility to understand what is happening in the local media environment You need to understand: It's a bit difficult to erase someone else's data, but if you can contribute to a generative AI tool, you can collect the media articles that are most relevant to you, have the tool summarize them, and give you answers. You can limit yourself to what you’re typing, and that can be very valuable and a huge time saver,” says Graviss.
“The spoken word is the currency of the State Department, so we can view generative AI as a true force multiplier…We want our diplomats to spend less time on screens and more time face-to-face.” ” Gravis said. . “We want to get them interested and spend less time writing about it, and this helps. Generative AI helps in a pretty significant way,” he added.
NIH uses AI to match research applications for funding with program personnel at the National Institute of General Medical Sciences. The institute receives more than 20,000 applications each year, but only about 4,000 receive funding.
“Matching applications to program personnel at the institute can be a very time-consuming process,” Greglick says.
The institute uses AI to capture title and body information in grant summaries and uses that information to route applications to the most appropriate program personnel.
“This allows us to narrow down the amount of work that our staff has to do to really determine, 'Is this application within the scientific expertise of this program person?'” Greglick says Mr.
Analysis center will “upskill” State Department employees
In his role, Mr. Gravis also leads the State Department's Center for Analytics, which serves as the central hub for all data and AI capabilities. The State Department recently stepped up data collection for diplomatic operations and appointed chief data officers in each department and office.
“We have adopted this campaign approach…taking one foreign policy issue and one management issue at a time. [and] Proceed as quickly as possible within 6 months, upskill people in the field, deliver value quickly, devise sustainable solutions, and the analytics center as a hub will focus on two priorities: It allows us to pivot,” Gravis said.
Over approximately two years, CfA plans to complete 16 campaigns in some of the department's most important areas, including strategic competition, workforce diversity, cybersecurity, and multilateralism.
“Each of these areas will be elevated through the use of data and AI,” Gravis said. “I think this is a model that has proven to be really successful,” Gravis said.
Copyright © 2024 Federal News Network. All rights reserved. This website is not directed to users within the European Economic Area.