What Impact Is AI Having in Government Research?

With the increased use of artificial intelligence (AI) in government research, questions regarding its impact on public policy and its ability to address today’s pressing challenges have naturally surfaced. Government agencies want AI methods that are transparent, understandable, and reliable. They want to know how these methods can best inform policy decisions and the government’s abundant forecasting needs. Here to explore how these methods—from chatbots and social media analytics to analyzing massive datasets—can contribute to federal research, public impact, and public engagement are Westat’s Elizabeth Mannshardt, PhD, Vice President and Director of Statistics and Data Science, and Gizem Korkmaz, PhD, Vice President of Data Science and AI.

Q. How is AI currently being used in federal research, public health projects, and in supporting public policy, and what impact has it had so far?

A. Gizem Korkmaz: AI tools help identify trends and topics of interest, understand public sentiments in real time, and also increase citizen engagement. AI has already been actively integrated into federal research, public health initiatives, and public policy across government agencies. In fact, as of January 2025, over 40 agencies submitted 2,000-plus applications of AI use, more than 270 of which were in the area of health and human services and 240 in the criminal justice space.

Liz Mannshardt: By making federal data more discoverable and accessible, AI expands its reach and supports greater public awareness and engagement. These tools also facilitate policymakers’ understanding of the impacts of policy decisions, enabling them to act more swiftly and effectively. AI’s ability to rapidly analyze massive datasets, for example, has enabled Westat to help the government forecast and mitigate the spread of infectious diseases and foodborne illness outbreaks. Westat has also applied AI to support government efforts in monitoring, preventing, and reducing adverse drug events (ADEs). Additionally, by developing methodologies to link data sources across organizations and generating synthetic data with AI, Westat has been able to enhance the utility of federal data while upholding strict standards for confidentiality and privacy.

Q. How have advanced AI techniques—such as transformer-based models and neural networks—improved data-driven decision-making for federal agencies and supported policy for the public good?

A. Gizem Korkmaz: Advanced natural language processing (NLP) and machine learning (ML) techniques like pre-trained domain-specific language models and deep learning are helping federal agencies in so many areas, ranging from epidemic forecasting and opioid crisis response to optimizing traffic flow, predicting crop yields, and detecting cybersecurity threats. In supporting the government’s rapid identification of ADEs, for example, Westat’s application of AI expands the government’s understanding of the indicators causing them and enables the prediction of future events. This involves the use of pre-trained large language models (LLMs) and ML to classify the severity of drug-related events and identify features predictive of severity. This work is critical for ensuring patient safety and informing regulatory decisions.

Q. How can federal agencies and the public ensure that AI-driven insights are transparent, explainable, and trustworthy, particularly in critical areas like public health and policy?

A. Liz Mannshardt: Human expertise and insights are paramount to building trustworthy and reliable AI capabilities. To ensure Westat AI-driven insights are clear, understandable, and reliable, we employ a system of checks and balances that includes subject matter and data science experts in the process, the use of “gold standard” data in our research and validation processes, and documentation of our methods. We also abide by the frameworks established by the National Institute of Standards and Technology (NIST), the Federal Committee on Statistical Methodology (FCSM), and the National Artificial Intelligence Research Resource (NAIRR).

Gizem Korkmaz: Given the guidelines and frameworks Liz mentioned, agencies and organizations need to ensure that these principles are implemented and enforced, and organizations are held accountable. AI governance is paramount to establishing this. In March 2024, the Office of Management and Budget (OMB) issued guidance requiring agencies to convene an AI governance body. Westat has formed an AI Governance team that is currently developing our AI governance strategy and protocol based on the guidelines and frameworks.

Q. What are some of the biggest challenges government agencies face in leveraging AI, and how can they overcome them?

A. Gizem Korkmaz: While AI-driven technologies can generate solutions to address critical global challenges, contractors and others working with government agencies need to be cautious in AI’s use. Humans are necessary in the loop with experts checking, intervening, and making the final decisions to ensure both ethical and the best outcomes. We also need to ensure that this technology and the necessary resources are accessible to all government agencies. The good news is that there are efforts underway through organizations such as the NAIRR to solve the access divide. Last but not least, agencies will need to gain AI literacy to understand AI’s strengths and limitations and how to effectively use it. Agencies will need to allocate resources specifically for training purposes.

Q. Looking ahead, how do you see advancements in AI, data science, and sustainable AI practices having a transformative impact on federal research and the public good in the next few years?

A. Liz Mannshardt: As AI and data science evolve exponentially, the next few years will not only bring significant improvements in data discoverability and accessibility but also emphasize the importance of sustainable AI practices—minimizing energy use, reducing carbon footprints, and responsibly managing computational resources. This combination will enable the public and policymakers alike to gain timely, actionable insights while ensuring that the innovative tools we build today don’t compromise our environmental future.

Our expert staff members stay on the cutting edge of AI methods—developing energy‑efficient models, leveraging carbon‑aware computing strategies, and optimizing algorithms to deliver robust performance with minimal environmental impact. By partnering closely with clients, we implement custom AI solutions that advance federal research objectives and enhance public‑good applications.

Gizem Korkmaz: We also expect greater collaboration among federal agencies and among experts—all contributing to public policies for the public good. The development and oversight of responsible AI is a team sport. It requires collaboration among diverse experts, including social scientists, data scientists, statisticians, computer scientists, IT professionals, and policymakers to develop, evaluate, and refine methods and outcomes.

Q. Any final thoughts about the future of AI?

A. Liz Mannshardt: AI methods and tools are evolving rapidly, opening up exciting new possibilities. As these technologies become more powerful—especially in areas that affect people’s lives—it’s essential to involve experts in subject matter, data science, and computer science to help guide their development and use. AI will be key in supporting policies and meaningful solutions for the public good.