Much of the dialogue in our industry has been dominated by Artificial Intelligence (AI) and what that means to the fields of Knowledge Management, Data Management, and Content Management. At EK, we’ve been witnessing firsthand the impact of the “AI Revolution,” with markedly greater inquiries from organizations seeking AI services or software advice, as well as clients experimenting with what the market currently has to offer.
The buzz is clearly around generative AI, with tools like ChatGPT, DALL-E, and Microsoft Copilot inspiring a great deal of interest and enthusiasm. We’re playing with these tools, and many others, and they demo spectacularly and are a lot of fun to experiment with. Though extremely promising, however, there’s still a major gap between a good demo and a tool that can be trusted at the enterprise level to deliver not just the right answers, but complete answers that fully leverage the knowledge and expertise of the organization and its people.
Where generative AI tools fall short is in their prevalence for hallucinations (as in, a rather gentle word for errors). Imagine a situation where the AI of a medical supply company recommends the wrong product, where an airline offers a steeply discounted ticket, or where a leading services firm’s chatbot insults a potential customer. If you’ve followed the news, you know each of these issues (and many more) have already occurred. The risk for errors is steep, and with those errors can come unacceptable risks that can lead to loss of revenues, regulatory fines, brand and reputation issues, or even loss of life.
Moreover, the current slew of tools are capable of understanding structured content and unstructured content increasingly well, but struggle to deal with conflicting inputs, and outright fail to incorporate an organization’s most valuable assets, its tacit knowledge and human expertise. This results in the classic issues of “garbage in, garbage out,” and worse yet, fails to connect people and their knowledge in any actionable form.
To reframe the conversation, we’ve been asking our clients to consider what they’re really seeking to get out of AI. We find that most organizations are seeking the promise of AI, but require the reliability and trustworthiness of more conventional knowledge assets, and moreover are seeking to incorporate human intelligence into their solutions in a way that captures all of an organization’s knowledge assets, rather than leaving key elements like people, their experiences, and their expertise out of the equation. To that end, we discuss Knowledge Intelligence (KI), rather than AI.
KI integrates institutional knowledge, business context, and expertise to enhance AI capabilities through effective knowledge and data management practices. Put simply, it smashes together the latest in AI technologies with all of an organization’s knowledge assets, delivering on the promises of AI with greater reliability and organizational expertise.
The specific components of KI address the following critical gaps for AI:
- Expert Knowledge Capture & Transfer: Programmatically encoding expert knowledge and business context in structured data & AI;
- Knowledge Extraction: Federated connection and aggregation of organizational knowledge assets (unstructured, structured, and semi-structured sources) for knowledge extraction; and
- Business Context Embedding: Through providing standardized meaning and context to data and all knowledge assets in a machine-readable format.
In a way, this feels like the moment in Knowledge Management that I’ve been waiting for. With over a quarter-century of KM consulting experience behind me, I’ve been talking about the potential for these use cases and business outcomes for some time. We’re now in a position to harness all of the value of KM for enterprise-level outcomes that can be deeply transformative for an organization and impactful for its employees, partners, and other stakeholders.
So what does this mean? What can KI really do? First, it incorporates people and their true expert knowledge into the AI capabilities of an organization. If you’re lacing together all of an organization’s assets, the lessons learned, how-to’s, and “mistakes made” of an organization’s experts are some of the most valuable. KI ensures they’re wired into the complete body of knowledge, information, and data, incorporating all assets to return comprehensive answers.
Next, KI brings both quality and context into the equation, eliminating (or at least drastically decreasing) the hallucinations and errors by ensuring the system ingests accurate and actionable information, and by adding taxonomies and ontologies to guide the model and improve the ability for accurate inference and connections.
Further, KI is explainable. Too much generative AI is a “black box,” meaning you get an answer, but you don’t know from where the answer came, or what logic went into deriving it. KI is explainable AI, showing sources and revealing any inferences, or assumptions, that were made in deriving the answer.
All of this translates into what most organizations really mean when they say, “we want AI,” but in a way that is reliable, trustworthy, understandable, and consistent. Are you ready to move beyond the AI pilots and demos and realize something of real value for your organization? If so, we’ve been doing this for a decade and stand ready to help you achieve KI.