AI can read sentiment. But can it understand organizational culture?
I think that distinction matters a lot when we talk about AI creating workplace training. Sure, feed it enough survey comments, employee feedback, emails, discussion posts, and other organizational data, and it can identify patterns to beat the band. It can tell us that people are frustrated with a new system, employees don't feel supported, or managers are worried about AI. It can summarize those signals and make useful inferences from them.
But culture is something different*. It's what people know about how things really work around here and who really has influence, regardless of the org chart. It's what you can safely say in a meeting and what really gets rewarded. Culture includes the stories people tell about a past technology rollout, and it knows which groups are quietly resisting an initiative and why.
Often all that is more tacit than explicit. Networks, communities of practice, showing your work, and culture are the places where people exchange information, develop expertise, build trust, negotiate meaning, and figure out how work gets done. The elements of social infrastructure create context, where people interpret what's happening and decide what to do about it.
AI can analyze some of what goes on in the social infrastructure, and can even produce a remarkably convincing description of an organization's culture. But a description of culture isn't necessarily an understanding of culture. Really: Go look at the sacred story told by your organization's marketing materials, and compare it to what you know to be reality. (I am reminded of the organizations that solicit "innovative, out-of-the-box thinkers", then hire people who've never been in trouble.)
My graduate work involved a deep dive into first- and second-order barriers. While AI might tell you that staff need more help with "understanding" AI and cranks out more "how to use AI courses", culture knows that there are other issues: employees don't trust leadership because of previous technology initiatives; managers say they support experimentation but punish mistakes; people have learned that admitting they don't understand something is career-limiting. Or maybe the workers have learned to just wait the thing out until the next initiative comes along.
These aren't "content" problems, but social and organizational ones.
That's why I don't think the future of L&D is just about getting better at prompting AI to create courses. The more interesting work may be helping organizations understand the human and social conditions that determine whether those courses will actually help with performance problems.
L&D's value isn't just in producing learning content; it's in helping organizations interpret, connect, and act within the social infrastructure of work.
*Some material from Bozarth, J. "Show Your Work". Wiley/ATD
Posts
