Accenture Federal Services explores generative AI, government applications through new center

The federal government stands to get even cozier with AI-driven technologies like ChatGPT.

Arlington, Virginia-based government consulting firm Accenture Federal Services (AFS) recently launched its Federal Generative AI Center of Excellence. Generative AI creates text, images and more in response to a prompt; it most recently made headlines with the rise of ChatGPT. AFS said that the center will act as a design hub for prototyping and scaling, host accelerators and generally explore how generative AI can help tackle challenges federal agencies face.

The center will work closely with the Generative AI and Large Language Model Center of Excellence, developed by AFS parent company Accenture. AFS’s center will work also work with both commercial and public sector clients.

Michael Thieme, generative AI lead at AFS, told Technical.ly that the humanistic nature of generative AI makes it so interesting and special for government work.

“What’s newest and most disruptive about [generative AI] is that it’s closest to the human,” Thieme said. “Meaning that as opposed to a technology working completely behind the scenes solving some predictive problem on some dataset you can’t see, generative AI is only useful when there’s a person typing into a keyboard asking for something to happen.”

Thieme said that work at AFS is very data-oriented, which often requires collaboration with data scientists, human interface designers and other experts. With the center, AFS wanted to bring everyone to one virtual location to understand a given piece of technology or issue — and not have 20 different people solving a problem separately.

“Generative AI has captured the world’s attention faster than any other recent technology,” said John Goodman, AFS CEO. “But how people are using it right now is just the beginning. We can already see that this technology is poised to shape the future of science, enterprise data, and how we design and deliver solutions for federal missions.”

Thieme envisions several uses for the center, including unlocking generative AI’s potential in data analysis. For instance, a client could have 10 years of PDFs on intelligence analysis archived; a normal tech response would involve annotating and tagging the data in an effective, useable format and developing output summaries. But generative AI already has conversational potential — that is, asking what AI can tell you about a data set and building on questions already asked.

“You get back a continuous answer as opposed to having to go back to the drawing board, ask one question a thousand times or build out that conversational capability, it allows you to interact with your data in a very human-oriented way,” Thieme said.