RAG Could Unlock AI’s Human Potential in Government
At the AI in Action Workshop in Washington, D.C., OpenAI Partnership Manager Alex Bonnell highlighted the potential of retrieval-augmented generation (RAG) to transform government use of AI by combining large language models with external knowledge bases to create more accurate, context-aware insights. She emphasized that RAG empowers individuals and teams to curate personalized “knowledge universes,” breaking down silos and enabling employees to leverage collective expertise. Drawing on her experience as CIO of the Air Force Research Laboratory, Bonnell noted RAG helped reduce IT tool sprawl from 183 systems to 47, underscoring its efficiency benefits. She described AI as the first “intimate technology,” allowing users to engage with it on personal terms, and outlined four stages of AI adoption: exploration, overcoming fears, recognizing efficiency gains (while overcoming guilt), and routine integration into work. While stressing AI should be used selectively and intentionally, Bonnell concluded that its impact depends less on technological speed and more on how individuals choose to engage with it—whether as a tool for growth or as a shortcut.