In This Story
In the labyrinth of routine government operations, from immigration processing to climate change mitigation, there exists a common critique: the sluggish pace of the U.S. federal government.
A new multi-university study led by Schar School of Policy and Government PhD graduate and postdoctoral faculty member Muhammad Salar Khan suggests a transformative solution lies in leveraging Artificial Intelligence (AI) technologies in the public sector.
Khan is a Don E. Kash Postdoctoral Fellow in Science and Technology Policy whose coauthors include climate change professional Azka Shoaib and Virginia Tech PhD candidate Elizabeth Arledge.
“As both a public policy student and scholar, I’m always intrigued by how we can use various AI algorithms to design tax policies that enhance economic efficiency and equity, reduce immigration backlogs, improve communication within federal government agencies, and achieve many other social policy goals,” Khan said. “With all this in mind, I was motivated to conduct the AI study.”
Smart homes, autonomous vehicles, and AI-driven services such as delivery drones, GPS navigation, and virtual assistants like Amazon’s Alexa illustrate the capabilities of AI technologies, not to mention generative AI platforms like OpenAI’s ChatGPT and Google’s Gemini, technology that is reshaping engagement and information processing across sectors.
The adoption of such AI technologies by the federal government could significantly enhance productivity and efficiency.
However, the integration of AI into federal operations is fraught with challenges, say the study’s authors. Ethical and legal concerns, antiquated infrastructures, a workforce unprepared for the transition, institutional resistance, and a general lack of public acceptance loom as significant barriers.
“We found that AI usage is heavily concentrated in a small number of federal agencies,” Khan said. “AI technologies are in varied stages of development, and the sophistication levels of these technologies vary from simple linear regression to more complex machine learning methods, such as deep learning models.”
Among the surprises Khan and his team discovered included the lack of effective evaluation of progress on responsible AI adoption and implementation in both legislative and executive branches, highlighting the need to identify opportunities for improvement.
Addressing these issues necessitates a robust policy strategy that intertwines scientific innovation, policy reform, and economic incentives. In a comprehensive review of policy frameworks, the study’s researchers sought to chart a course for AI adoption in government practices.
The findings from these frameworks could pave the way for a set of actionable policies that promote the integration of AI functionalities within the U.S. federal government, potentially transforming how government services are delivered and enhancing their impact on societal challenges.
The study urges policymakers to consider these insights as they formulate strategies to overcome the hurdles to AI adoption and harness its potential for public good.
“Now that the study is published, I hope that it sparks further discussion and action in the realm of AI adoption within the U.S. federal government,” Khan said. “Specifically, I hope that policy makers and relevant stakeholders take note of the findings and recommendations outlined in the study and begin to implement necessary changes to promote responsible AI adoption.
“This could include initiatives to streamline the AI landscape, improve evaluation processes, incentivize collaboration among agencies, and increase funding to make AI models and decisions more interpretable and transparent.”
In the end, “ultimately, I hope that the study contributes to the advancement of AI policy and practice in a way that benefits society as a whole,” he said.