|
3. GPO's Model Context Protocol
Rep. Mike Carey (OH-15): Chairwoman Bice and I are constantly looking to understand how other government offices are incorporating AI and machine learning systems into their daily operations. Which brings me to the GPO's Model Context Protocol or the MCP. The Model Context Protocol is an open source standard that allows connecting artificial intelligence applications to external systems. So in short, the public can use this protocol to examine the troves of data and better understand government information. Can you share a little bit more about this development and as you share that, maybe explain in layman's terms so people can understand exactly what that what as you're describing, what you're doing, what it means.
The Honorable Hugh Halpern: Sure. So the Model Context Protocol, and I will preface this with I am not an expert in constructing APIs and these similar kinds of things, but I'll do my best. The Model Context Protocol is an avenue in for these other kinds of LLMs and other kinds of AI models so that they can utilize our repository of data and use it without having to constantly train on it ahead of time. So it's a lot more efficient, particularly for a data set that changes literally every day, multiple times a day. Every time we upload a new edition of the congressional record or a new group of bills that had been introduced the day before, that data set changes. But this enables us, this enables your AI tool, whether it's Claude or Gemini or ChatGPT or whatever, it gives those tools a route into our data set so that it can effectively query that data set, find the answers it's looking for, and then present that back. Everybody was talking about, at the beginning of sort of this AI revolution, what chatbot are we going to adopt for GovInfo? And that's not the best use of our resources, because folks on the outside are going to be a lot smarter about this than we can be.
|