Community First: The Socio-Technical Key to AI Adoption in Higher Education
Observations from five communities on how people actually adopt AI.
Last week, I watched a cohort of staff present their final projects in the AI@Work Certificate Program at UC Santa Barbara. Ten weeks of structured learning. Real problems. Real tools. Real deadlines. By the end, every participant had learned far more than they knew at the beginning. But the most striking thing was not what they had learned individually. It was how they had learned it—together, as a community teaching itself.
One participant credited a colleague for teaching her Python in the sandbox environment. She had never written a line of code before. Another said she got “so many more ideas” from watching a peer’s presentation than she had generated on her own. A staff member who had hit every possible roadblock—failed prompts, hallucinated outputs, broken workflows—summed it up in five words: “I haven’t given up.”
Nobody presented a flawless product. Everybody presented a real journey. People shared what went wrong as openly as what went right. They named the colleagues who helped them. They asked each other questions. They offered to keep meeting after the program ended.
The enablement was not the tools or the curriculum. It was the community they built while learning together.
AI is socio-technical. Our adoption strategies should be too.
AI systems are socio-technical. The NIST AI Risk Management Framework states it directly: AI systems “are influenced by societal dynamics and human behavior.” The OECD AI Principles, ISO standards, and the EU AI Act all build on the same premise. The technology does not operate in isolation. It operates in the context of people, workflows, organizations, and culture.
If AI is socio-technical, then our approach to adoption must reflect both sides. In my observation, the technical side—tool selection, platform procurement, licensing, infrastructure—receives more structured attention. The social side—how people learn together, build shared understanding, and develop the confidence to change how they work—often receives less deliberate investment.
Community is that social side. And in my experience, it is the side that determines whether adoption actually happens.
What community actually does for adoption
Every conversation about AI adoption arrives at the same question: How do we get people to actually use these tools? Typical answers include training, documentation, champions, and executive sponsorship. Those matter. But what I have observed over the past two years, through communities I have helped build, lead, and participate in at the campus, system, and national level, is that the mechanism that moves people from awareness to adoption is community.
Not community as a buzzword. Community as a specific, observable thing: a group of people who share a context, learn in the presence of others, and hold each other accountable for trying.
Here is what community does that training alone does not:
It reduces the stigma of failure. When someone shares a failed prompt in front of peers, it gives permission for everyone else to struggle openly. People did not hide their dead ends. They narrated them. Every time someone described a failure, others leaned in. That is how risk tolerance spreads.
It surfaces domain knowledge. Every participant started with a problem from their own work. What made the presentations valuable was not the AI. It was the domain expertise. AI was the catalyst. The knowledge exchange was the product.
It creates shared vocabulary. After ten weeks, the cohort could talk about prompting, grounding, hallucination, and human-in-the-loop review as shared concepts. That shared language is what lets people communicate about AI across departments.
It sustains momentum. Individual training fades. Community provides the follow-up that training cannot—someone to ask when you are stuck, someone who tried the same thing and found a workaround.
It produces relationships and networks no one planned for. Unexpected connections form across departments and open doors long after the program ends.
Communities at every scale
This dynamic works at every scale. As consultant to the AI@Work Certificate Program, co-director of the UCSB AI Community of Practice, co-founder of the IT Professionals Mentorship Program, co-lead of the UC AI Community of Practice, and co-lead of the EDUCAUSE AI Community Group, I’ve seen the same mechanism in action.
The certificate program is formal: structured curriculum, milestones, presentations, and accountability. The UCSB AI Community of Practice is informal: voluntary, no grades, driven by curiosity and contribution. The IT Professionals Mentorship Program builds the mentoring and collaboration skills AI adoption requires. At the UC system level and through EDUCAUSE, communities connect practitioners across institutions so successes (and lessons) travel faster.
Formal structures create accountability. Informal communities create permission. The strongest approach is when both exist and feed each other.
What the presentations taught me
Most AI strategies focus on tools and policies—procurement, licensing, acceptable use guidelines. Those are necessary. They are also insufficient.
Deploying tools without investing in the people and communities to support them is not a strategy. It is a hope.
The presentations showed me something different. Tools gave capability. Community gave confidence, shared vocabulary, and people to turn to when things didn’t work. Domain knowledge and human judgment remained central. When AI succeeded, the time savings were massive. When it didn’t, the community kept people going.
If you are building an AI strategy, start with community. The relationships, trust, and shared understanding you create become a lasting foundation for future initiatives—well beyond any single tool or project. The tools will follow. The people need each other more than they need another platform.