AI and Knowledge Management: How KM Will Change
As a complement to my long reflection on KMWorld 2023, I offer this more practical set of insights on specific ways KM will likely evolve with AI integrated into KM solutions, used to analyze data and prepare it for learning, and offer complementary structures for knowledge representation. This is an initial pass at the most important components of knowledge management—the notes on changes will be updated as more evidence becomes available from research and client interactions.
AI and Knowledge Management
| KM Concept | AI and Knowledge Management: How AI Will Change KM |
|---|---|
| Knowledge Capture | AI will do very little to change the capture of knowledge, save for automating transcripts from verbal conversations. |
| Knowledge Curation | AI will summarize content and identify categories and tags, but it will do little to help with change management of knowledge impacted by changes in context, such as new technologies or social norms. Curating AI will likely become a new KM task as organizations train their own models, integrate bespoke knowledge graphs, and otherwise seek to transfer their knowledge to LLMs and related technologies. |
| Connections and Retrieval | AI will connect people to content in new ways, and it may well connect people, or at least introduce them, if security and privacy models allow for rich representations of people, their work, their skills and their interests. AI will likely play a significant role as the next primary interface for retrieving enterprise knowledge, replacing traditional search with more proactive retrieval options. Many new retrieval engines will likely emerge, forcing organizations to rethink their knowledge management infrastructure. Because of its unique role in proactive contextual retrieval, AI will also increase the opportunities for knowledge reuse. |
| Collaboration | AI is unlikely to have a major impact on the act of collaborating. In some cases, the output from models may be the object of collaborative work, and research related to that work may also be discovered and brought to the attention of the collaborative group, but the dialog, feedback, revisions, and other activities, while informed by AI will not likely find it as a participant. |
| Knowledge Creation | Generative AI’s primary role is to create, but it creates from existing knowledge in most cases rather than net-new knowledge. Creating knowledge will remain primarily human, although, in some areas such as design, chemistry, and engineering, some forms of generative AI may offer synthesized views that recombine existing knowledge into novel forms that, upon reflection, may be adopted by people as being acceptable. Because generative AI responses are ephemeral, knowledge managers will need to curate such responses as they may not be generated again if given the same prompt. |
| Knowledge Retention | Knowledge retention becomes tricky given the incompleteness, nuance and subtlety of written and recorded records. Even if digitalized knowledge is retained, in many circumstances, that knowledge is very different from what people actually know. Just because “knowledge” has been captured and can be retrieved is not the same as retaining the knowledge collectively owned by those who created, experienced, and continue to augment the digital expertise. |
| Sharing and Recommendations | Explicit sharing, for instance, clicking on a sharing button, will not be automated, though the suggestion of where and to whom the sharing will take place will benefit from AI. The act of explicit sharing may wane over time as generative AI agents learn to effectively make people aware of relevant content without needing a human to point it out. |
| Knowledge-based innovation | Knowledge-based innovation may use generative AI-based outputs to inform the innovation work, and it may even offer innovative takes on the subject as starting points—and perhaps provide alternative views on promising outcomes depending on the domain. |
More on AI and Knowledge Management: Summarization, knowledge organization and too much AI
AI will also support general horizontal applications, such as content summarization, that can be applied to codified knowledge. It may significantly influence areas like knowledge organization, making it more granular and specific but potentially more abstract and less transparent. If, on balance, the retrieval of knowledge improves in quality, how the knowledge is organized may no longer matter, impacting disciplines like taxonomy and tagging.
Organizations will likely find AI-based knowledge management options coming from multiple directions: within enterprise apps, as standalone options, within KM-specific applications, and through public tools like browsers and “AI” sites. Enterprises will likely face feature duplication and overlap. As organizations roll out AI-based KM solutions, they must rationalize their architectures to optimize KM features. Unlike collaboration, which can confuse people with where to collaborate, reducing productivity, too many AI responses can lead to confusion about the best answer among multiple responses.
KM disciplines will also prove valuable to organizations building their own AI models, as the management of the models will become a new form of represented knowledge that needs to be management.
For more serious insights on knowledge management, click here.
Thoughts about AI and Knowledge Management: How KM Will Change? Please leave a comment or ask a question.
Leave a Reply