Moving beyond the illusion of ‘AI is magic’ at KM & AI Summit 2025
Understanding AI as a true technological mechanism—not just the mystical ingredient that makes all your problems disappear—is fundamental.
At this year’s KM & AI Summit, Seth Earley, CEO of Earley Information Science and author of The AI Powered Enterprise, will break down what makes AI work—and work successfully—in his session, “Building a Knowledge Architecture for Retrieval-Augmented Generation (RAG).”
Unlock the future of innovation and knowledge management at the KM & AI Summit in Scottsdale, Arizona, March 17-19, 2025.
The KM & AI Summit—brought to you by the organizers of KMWorld—brings together thought leaders, technologists, KM practitioners, and industry experts to examine the latest trends and best practices for striking the balance between robust KM and AI innovation.
Earley explained that, at its core, “AI is not magic. Although it seems to be magic, we do need data. We need a foundation of good data, quality data—and we need the right data architecture, because at the end of the day, it's a retrieval mechanism.”
With a mechanism that depends on information—and good information, at that—developing a system that feeds it quality data seems rather obvious. But for the AI era, returning to and refining the basics of data and information management is less apparent, though more crucial, than one might think.
“There's a lot of what I'll call, aspirational functionality, where people are expecting certain things or trying to do certain things, but they're little ahead of their skis,” said Earley. “I still see vendors…assume there's a knowledgebase…[but] how good is that knowledgebase? How well curated is it? How well organized is it tagged? Is it fresh?”
With retrieval-augmented generation, or RAG—a popular method of enhancing AI outputs with external information—these questions ring even more true.
“When people say, just use generative AI, there's a whole retrieval piece of this. And the problem with retrieval is the problem that we've always had with retrieval, the problem we've always had with search,” said Earley. “Anything we're doing with retrieval-based AI is to make up for our past sins in poor data, content curation, and hygiene.”
“The way to do that is to build a knowledge architecture, [or] a subset of the overall information architecture of the enterprise, and it's focused on very specific use cases. It's a much narrower domain, and it's used to access information for the user in the context in which they're using that information. So, knowledge architecture for generative AI is absolutely essential to making retrieval-augmented generation work, full stop,” Earley continued.
Outside of his session, Earley looks forward to diving into the nitty-gritty of the real-world problems associated with KM and AI.
“I'm looking forward to seeing the real, hard-hitting, deep-dive types of case studies and examples,” he noted.
Earley further predicted that many attendees will be interested in effective resource allocation; as demands grow more ambitious and budgets get tighter, managing resources is more important than ever.
“It's all about resource allocation and course corrections. You have lots of opportunities to apply these tools; there's lots of initiatives going on. How do you decide what the best use of your resources are?” Earley pointed out. “I think anything that is helping people make decisions about how they apply their resourcing is going to be important, because it's really saying, how are we measuring value, how are we measuring outcomes?”
Earley’s session will take place on Tuesday, March 18 at 3:30 p.m.
For more information on KM & AI Summit and to register, please go to https://www.kmworld.com/KMAISummit/2025/Default.aspx.