The Federal Deposit Insurance Corporation (FDIC) is testing two generative artificial intelligence (AI) chatbots as part of a new pilot effort aimed at improving internal efficiency, the agency’s chief information officer (CIO) said Thursday.
Speaking at a Jan. 22 event hosted by the Digital Government Institute, FDIC CIO and Chief Privacy Officer Sylvia Burns said the agency launched the chatbots late last year to explore how GenAI could support basic research and workflow automation inside the agency.
Burns said the chatbots focus on generic use cases already within FDIC’s existing technology inventory, including automating routine tasks and assisting with internal research. The goal, she said, is to free up employees’ time for higher-value work.
“FDIC is a very risk-averse agency. It’s an agency that is very cautious about adopting new things,” Burns said. “The downside of that is that we have human beings still doing things very manually, which is not necessary.”
“So, I think the introduction of these tools is giving an opportunity for us to use our human capital and intellectual capital in a much more high-value way,” she added.
The pilots are also serving as a learning exercise for the agency. Burns said the FDIC has established daily office hours for pilot participants and internal discussion threads to surface questions, challenges, and lessons learned as employees interact with the AI chatbots.
For example, Burns pointed to an internal discussion thread led by an FDIC “AI guru” about how chatbot answers can change over the course of an extended interaction.
“They were talking about how when you start the prompt, the response will be truer to what you asked, but the more you have a conversation with the chat, it can diverge. The responses start to stray from … what your original ask was,” Burns explained.
“So, I just found that fascinating. I’m learning,” she said.
For now, Burns said the two chatbots have limited use cases. However, she predicted the tools will “really be huge” in terms of letting FDIC employees “focus our brain power on the hard things that the AIs can’t do.”