There’s a lot of speculation about where knowledge management is headed, most of it untethered from actual data. What the current research shows is more specific, and more useful, than the general “AI will change everything” narrative circulating in most commentary. Three forces are converging at once: a workforce retiring faster than knowledge is being captured, AI adoption outpacing governance, and organizations starting to recognize that KM maturity has been measured wrong for years. Here’s what the research actually says about each.
The knowledge loss problem is worse than most organizations realize
APQC’s research into what it calls the “Great Retirement” found that only 8 percent of organizations consistently capture knowledge from departing retirees, while 16 percent make no attempt to capture it at all. That means the large majority of organizations sit somewhere in between, capturing knowledge inconsistently, usually depending on whether a particular manager happened to prioritize it before someone walked out the door.
This is not a future risk. It is a current, ongoing loss, and it’s compounding. Every experienced employee who leaves without a structured knowledge transfer process takes tacit knowledge, the kind never written into a manual, with them permanently. Research from APQC‘s 2026 outlook frames this as one of the defining pressures on the field this year, noting that the AI fluency gap is here to stay, and organizations must upskill employees and capture institutional knowledge before retirements and turnover erode critical expertise further.
AI adoption is real, but most implementations are failing
The adoption numbers are genuinely high. McKinsey’s 2025 State of AI survey found that 88 percent of respondents said their organizations used AI in at least one business function. But adoption and success are not the same thing. Roughly 80 percent of AI projects fail to deliver their intended outcomes, and only about 30 percent of AI pilots progress beyond the pilot stage.
This gap between adoption and results is the central story in KM research right now, and it points squarely back to knowledge management fundamentals rather than technology limitations. APQC’s applied research on AI in KM is direct about where the actual barrier sits: the top barriers to AI adoption are rooted in governance and content management, including concerns about accuracy, data privacy, and compliance, rather than cultural resistance alone. In other words, organizations aren’t struggling to get employees to use AI. They’re struggling because the knowledge underneath the AI wasn’t governed well enough to trust.
Governance has moved from optional to foundational
For years, governance was treated as the unglamorous back-office part of KM, the piece that got deprioritized in favor of tools and adoption metrics. The research now places it at the center. APQC’s applied guidance is explicit that effective governance requires clear accountability for content validation, lifecycle management to prevent outdated content from training AI systems, defined roles for subject matter experts and KM liaisons, and oversight bodies such as steering committees or communities of practice.
Taxonomy, long treated as a housekeeping detail, has also resurfaced as a structural necessity rather than a nice-to-have, since a well-designed taxonomy enables AI systems to understand relationships between topics, domains, and business functions, which directly improves search, retrieval, and relevance. Without it, AI systems retrieve content but can’t meaningfully relate it to context, which is a large part of why AI-generated answers so often feel technically correct but practically useless.
Where organizations are actually finding success
The research isn’t only cautionary. Where AI-enabled KM is working, a consistent pattern shows up: knowledge quality work happens before the AI layer, not after. APQC’s account of Novartis’s KM program is a useful case in point, describing how the KM team centralized and curated trusted knowledge before layering generative AI capabilities on top, with strong governance and collaboration with legal and compliance helping ensure reliability and adoption.
This ordering, curation first, AI second, is the throughline across nearly all the research on what separates the roughly 30 percent of successful AI pilots from the majority that stall.
What organizations are prioritizing going into the rest of 2026
APQC’s most recent priorities research names the shift plainly: KM is moving from a support function toward core organizational infrastructure. Current priorities center on incorporating AI and generative technologies responsibly, identifying and prioritizing critical knowledge, increasing KM maturity through stronger governance and foundations, and enabling collaboration and participation across teams.
Notably, “governance and foundations” now sits alongside AI adoption as a co-equal priority, not a follow-up task. That’s a meaningful shift from how KM priorities were framed even two or three years ago, when technology adoption tended to dominate the conversation on its own.
What this means going forward
The research converges on a single, fairly unglamorous conclusion: the future of knowledge management is not primarily about which AI tools organizations adopt. It’s about whether the underlying discipline, capturing knowledge before it walks out the door, governing content with real accountability, and building taxonomy that gives AI something structured to work with, gets built first. Organizations skipping that groundwork are the ones showing up in the 80 percent failure statistic. Organizations doing it are the ones the case studies get written about.
Sources referenced: APQC’s 2026 Knowledge Management Priorities and Trends research, APQC’s Great Retirement research, APQC’s applied blog on AI in KM (Feb 2026), and McKinsey’s 2025 State of AI survey.