KMWorld 2024 Is Nov. 18-21 in Washington, DC. Register now for $100 off!

  • February 12, 2024
  • By Marydee Ojala Editor in Chief, KMWorld, Conference Program Director, Information Today, Inc.
  • Features

Promises and Perils of AI for Enterprise Knowledge Management

It has long been a conundrum facing enterprises of all types: How best to quickly discover accurate information that is stored (somewhere) within the organization. It’s there, you just know it’s there, but where? In the last few years, the emergence of generative AI has shown promise for alleviating, if not completely solving, this problem. Pinning their hopes on AI for comprehensive information discovery, enterprises are also well aware of potential pitfalls posed by the technology.

This important survey by Pryon, conducted by Unisphere Research, points to the tension created by data silos, legacy systems, budget restraints, and lack of management support. At times attempts to deliver the backbone for a truly informed enterprise seems like one of those jigsaw puzzles that became so popular during pandemic lockdowns. They are intricate, seemingly designed to frustrate, yet challenging but not so challenging that you throw the pieces on the floor in frustration.

The survey reveals that, while organizations are excited about the potential of AI to improve knowledge management along with information access, they are also mindful of potential risks regarding data quality, security, privacy, accuracy, and governance. Not new issues, certainly, to knowledge managers, but they are receiving added scrutiny once AI becomes involved. How well does technology deal with these very important issues and how much human intervention and oversight will be required?


As the survey reveals, AI is top of mind for organization but is still in the early stages of adoption. Plans are in the works to implement some form of AI in the next year. A major initiative, when organizations are first starting to put the puzzle pieces together, is the summarization and information retrieval from unstructured data. We’ve long known that unstructured data forms the bulk of what enterprises have stored away and that unstructured data is still hard to find for end users. Increasingly, too, expectations are that information will be findable in very short order and the time people are willing to spend looking for relevant data is shrinking. The hope is that AI will help meet those expectations and decrease time spent frantically—and sometimes fruitlessly—searching for information. In turn, that should boost productivity and improve knowledge sharing.

The current situation with multiple data silos and systems that can’t be easily cross searched is not conducive to a truly informed enterprise. Silos arise for a number of reasons. It might be a competitive rather than collaborative company culture. It might be individuals simply wanting to keep their data close at hand in a format they like. It might result from incompatible infrastructures. It might be lack of communication among teams or departments about what data they collect and retain.

Multiple systems that don’t talk to each other could be the result of mergers and acquisitions, where legacy systems are not brought into common compliance with internal standards. They could also have simply been developed as “shadow IT” projects, where someone creates a system without bothering to tell IT what they are doing. Why these data silos and competing systems came into existence can help guide IT staff and knowledge managers into eliminating them. It can also point to silos that have a valid reason for not being accessible to everyone, such as human resources, strategic planning, and some financial silos.

See the Survey HERE.

KMWorld Covers
for qualified subscribers
Subscribe Now Current Issue Past Issues