By Holly Zanville, Founder/Executive Director, Learn & Work Ecosystem Library

Five years ago, the Learn & Work Ecosystem Library began as an experiment. Could we create a neutral, publicly accessible collection of information to help people better understand the learn-and-work ecosystem?

At the time, the challenge was finding information. Important initiatives, organizations, reports, and emerging concepts were scattered across hundreds of websites and publications. We believed there was value in bringing them together, organizing them, connecting them, and making them easier to discover.

The Library now contains more than 2,300 knowledge resources, including glossary terms, organizations, initiatives, topic briefs, a cartoon gallery, and other curated materials. Every month, new content is added and the database continues to expand. 

However, as content changes, organizations transition, terminology evolves, and initiative conclude, the Library has a new challenge: How do you keep an ever-expanding knowledge collection current?

Our technology team recently introduced me to a term I had not previously used in this context: staleness.

“You mean like in stale food?” 

This term made me think of the lettuce sitting too long in my refrigerator, or leftovers hidden in the back of a shelf. If ignored long enough, they become unusable.

“Yes, that kind of stale,” they said.

Understanding that term in the context of the Library collection made me realize that knowledge collections face a similar problem. A glossary term may still be largely correct but miss important developments. An organization description may no longer reflect its current mission. An initiative may have ended years ago or added new partners along the way. A topic brief may not include the latest policy changes, research findings, or emerging practices. 

The content has not necessarily become wrong; it has become stale.

Historically, the solution was straightforward – assign people to continuously monitor, review, and update information. Large organizations can sometimes do this, but smaller organizations often cannot. Labor-intensive efforts can be costly.

But the alternatives are not appealing.  Stop adding content? Archive large portions of the collection? Periodically throw everything out and start over? None of those options seem wise.

Instead, we’re exploring a different possibility: What if artificial intelligence could help us identify content that requires attention before it becomes stale? The goal isn’t to have AI rewrite everything wholecloth or have it replace human editors entirely. Instead, it would help us monitor our growing collection and direct human attention where it is needed most.

In other words, what if AI could help us clean the refrigerator?

That question has led us to launch a new research project in summer 2026: AI-Assisted Stewardship of the Learn & Work Ecosystem Library.  

The project explores whether AI can help a small editorial team (three part-time people) maintain the currency, accuracy, neutrality, and trustworthiness of a large and growing knowledge collection. Rather than treating AI as an autonomous content creator, we ‘re interested in whether it can function as a Library team member — helping identify resources that may need review, monitoring changes in the environment, and eventually suggesting updates for human consideration.

The project raises questions that extend well beyond our Library: 

• What happens when collections grow faster than humans can monitor them alone? 
• How should organizations balance automation and editorial judgment? 
• Can AI help maintain trusted knowledge resources without compromising quality or neutrality? 
• What level of human oversight is appropriate? 
• And perhaps most importantly: What does responsible AI-assisted stewardship look like in practice?

We do not yet know the answers. That’s why we’re conducting the research.

Over the coming months, our team will test new approaches, evaluate recommendations generated by AI, and document what works, what does not work, and what surprises us along the way.

The questions extend beyond the Learn & Work Ecosystem Library. Many organizations are wrestling with how to maintain large collections of information in an era when knowledge changes rapidly and resources are limited. We believe the lessons learned may be useful to a range of organizations — libraries, professional associations, research repositories, knowledge hubs, standards organizations, and others responsible for maintaining trusted information resources.

The challenge facing knowledge collections today is not simply creating information, but maintaining it. That may be where AI can help most.

Stay tuned.

See Proposal: Research Project: Exploring AI-Assisted Knowledge Stewardship: Maintaining Currency, Accuracy & Trustworthiness of Curated Knowledge Collections

See Project Page: here