IT Service Management: AI Doesn’t Replace Knowledge Management – It Demands It

You might think that the introduction of AI to IT support (and the wider ITSM and service management disciplines) creates a clean slate for your organization’s knowledge management initiatives. After all, generative AI (GenAI) can – through what seems like magic – provide answers to all manner of questions, offering up knowledge as needed.

However, it’s a misconception that AI replaces the need for knowledge management in IT support. As many organizations have learned the hard way, AI doesn’t eliminate the need for knowledge management, and, unfortunately, it amplifies the consequences of weak knowledge management.

This blog looks deeper into the link between AI and knowledge management to explain why AI doesn’t replace knowledge management and instead demands it.

Key Misinformed AI Assumptions Related to Knowledge Management

There are many misinformed assumptions around AI and knowledge management. A key one is that AI will automatically fix poor knowledge management practices. However, if your organization’s knowledge base is poorly structured, incomplete, or outdated, AI won’t magically fix it. It’ll just automate the mess.

Another is that AI-based knowledge management works the same way for all organizations. But, the ability to implement and scale AI solutions varies widely. For example, large organizations might have the technical expertise and resources to adapt and expand AI-enhanced knowledge management systems. In contrast, smaller or less digitally mature organizations might lack these resources.

Another is that AI-generated knowledge doesn’t need human validation. Not only should AI outputs be critically assessed (at least for now), but your organization must safeguard trust, fairness, and legal compliance in its use.

There are, of course, other misinformed assumptions, but hopefully these examples are sufficient to “sow the seeds” of concern about how AI and knowledge management are employed together.

The Impact of Weak Knowledge Management

So what happens if your organization adopts AI capabilities on top of a weak knowledge management foundation?

Common sense says that the AI learns from “bad data” and will automate what were manual mistakes at scale. For example, with IT support, AI-generated instructions might sound credible, but following them might not only cause frustration but also an adverse business impact.

But this is potentially just the proverbial “tip of the iceberg.” Many IT support AI use cases might struggle due to existing knowledge management issues, such as:

• Tickets getting routed incorrectly
• Ticket prioritization being based on flawed historical data
• “Hallucination” resolutions frustrating end-users (and wasting their time)

Then, because AI delivers incorrect results, we lose confidence (in AI). At best, your pilots might fail, but people – including leaders – will likely quite rightly start to question the ongoing investment in AI.

Ultimately, weak knowledge management will likely create an insurmountable barrier to your organization realizing value from its AI investments.

What’s Needed for AI Success

You can’t escape the calls for data quality as the foundation for AI success. Your AI investments will need:

• Clean, accurate, structured data
• Consistent categorization, tagging, and metadata
• Regular verification processes.

But this is just the beginning.

The need for AI governance has “loomed large” over organizations and ITSM teams for the last few years. Importantly, this isn’t just the operational “AI guardrails” included in ITSM tools, with traditional IT governance insufficient for AI adoption due to unique challenges, such as:

• Autonomous decision-making and what happens when things go wrong (and whether errors are detected in a timely manner)
• Lack of transparency, with this linking to trust in AI
• Security risks – with the organization liable for AI decisions
• Ethical issues, including bias caused by unsuitable training data
• Evolving regulatory requirements.

Then, a “knowledge architecture” is needed for AI consumption. This includes:

• Semantic search that understands intent
• Structured content
• Active metadata
• Integration with key business platforms
• Role-based access controls.

Ultimately, your AI investments need a solid knowledge management foundation.

How AI Improves Knowledge Management

The connectivity between AI and knowledge management is bidirectional. Knowledge management helps with AI success, and AI helps with knowledge management success. The latter includes:

• Content access and creation – AI can automatically find relevant knowledge base articles for agents and employees through contextual search, and draft new articles based on tickets that have been successfully solved. Importantly, AI eliminates the manual effort that previously made knowledge management unsustainable and powers faster ticket resolution.
• Knowledge organization – AI can monitor search patterns, identify knowledge gaps, and improve content performance through data-driven insights. It also enables dynamic knowledge management by curating, updating, and recommending relevant documentation in real time.
• Contextual knowledge delivery – AI can analyze user roles, behavior, and context to deliver relevant knowledge to the right person at the right time. This includes improved self-service capabilities, with AI offering solutions before users submit a ticket – thus reducing ticket volumes.

However, these capabilities and the associated benefits demand that your foundation is right – clean data, proper governance, and solid knowledge management to begin with.

It’s About More Than Technology

While it’s easy to focus on how the technology change will help IT support, there’s also a need for people change. For example, your staff’s roles might evolve from doers to designers, overseers, and analysts – with people becoming the architects of automated systems rather than being the manual processors of tickets. This change includes a greater appreciation of the importance of knowledge and knowledge management capabilities.

There’s also a mentality change required – from reactive to proactive thinking. Focusing on eliminating the root causes of incidents, not just resolving them faster, and using automation to prevent issues, not just speed up ticket closure.

Success measurement also needs to change – from ticket metrics to business outcomes. Again, knowledge management has a part to play in this redefining of “what good looks like” for IT support.

AI Doesn’t Replace Knowledge Management, But It Can Enhance It

What this all means: AI doesn’t replace knowledge management; when done right, it enhances it.

As a starting point, if your knowledge management is weak, AI won’t fix it. In fact, the introduction of AI on top of poor foundations could simply automate bad practices, scale errors, and erode trust in your ITSM team. If you expect AI to compensate for poor knowledge management, you’re setting your organization up for failure. AI needs a good knowledge management foundation to deliver the benefits your organization has been sold on.

Ultimately, AI raises the bar for knowledge management, exposing any weaknesses in the current approach. Put simply, AI doesn’t replace knowledge management – it demands it. So make sure your knowledge management capabilities are ready to enable the many benefits of AI.

Blog Post: How AI Will Influence Core ITSM Capabilities

Blog Post: The Human Touch in AI Adoption

Blog Post: The Skillsets IT Service Desk Agents and Managers Need in the Age of AI