As AI floods the internet with content, the humans who filter signal from noise are becoming invaluable. Here's how Liminary is building the platform to amplify their work.

We're living through a strange paradox. AI can now generate infinite content: articles, images, videos, research summaries. Yet finding information that's actually useful has never felt harder. The problem isn't scarcity anymore. It's abundance without meaning.

As media theorist Clay Shirky observed: "It's not information overload, it's filter failure." The filters we relied on (editorial gatekeepers, algorithmic feeds, search engines) are buckling under the weight of AI-generated content.

This is why human curation is experiencing a quiet renaissance. And it's why we built Collections on Liminary.

The Automation Trap

When ChatGPT launched, many predicted the death of the curator. If AI can generate infinite content, with AI-created articles surpassing human-written ones in volume, what's left for humans to do?

Duke University's Nasher Museum decided to find out. In 2023, they let ChatGPT curate an exhibition from their 14,000-object collection. The AI could select works, write descriptions, even propose themes. The result was technically competent but ultimately hollow.

As Chief Curator Marshall Price put it, AI proved useful for structured tasks: cataloging, basic descriptions, initial filtering. But the deeper work of curation (deciding why these pieces matter together, what story they tell, what unexpected connections illuminate something new) remained stubbornly human.

AI can process information. It can't tell you what's worth knowing.

What AI Enables (Not Replaces)

This points to a division of labor that's emerging across knowledge work.

AI handles the tedious parts: ingesting content in multiple formats, extracting key information, enabling natural language queries, surfacing connections you might have missed. This is genuinely valuable work that used to take hours.

Humans handle the judgment: deciding what belongs, providing context, making a collection coherent. Defining what makes something worth saving in the first place. Translating why a research paper means something different to a first-year PhD student than to a venture capitalist evaluating a deep tech startup.

As the World Economic Forum recently observed, "information no longer differentiates people—agency does." The abilities to ask better questions, navigate ambiguity, and turn ideas into action are the defining competencies of our era.

The best collections aren't just comprehensive. They're surprising. They juxtapose ideas that seem unrelated until a curator shows you why they belong together. That's not a task AI will automate. It's a task AI makes more valuable.

The Netflix Playbook for Knowledge

There's a reason Netflix succeeded in the age of infinite streaming content: when everything is available, the value shifts from access to curation. Netflix's recommendation engine, powered by both algorithms and human editors, became more valuable than any individual show.

The same dynamic is playing out in knowledge work. We're drowning in information at an estimated $1 trillion cost globally. What we lack are trusted guides who can tell us: here's what actually matters for this domain, and here's why.

When a human expert stakes their reputation on a collection, that means something. They can be questioned, challenged, engaged with. They have track records that can be evaluated over time.

The Published Collections Model

This is the problem we built Liminary to solve. Our core product captures knowledge across all the tools and contexts where people actually work: web pages, PDFs, YouTube videos, ChatGPT conversations, Gmail threads, local files, and uses AI to surface the right information at the right moment.

But we realized individual knowledge management was only half the opportunity. The other half: making it easy for experts to share their curated knowledge with others.

That's what Open Collections does.

An Open Collection is a curated library that anyone can browse and query. The curator saves content to their collection over time (articles, research papers, videos, documents) and Liminary's AI makes it searchable and conversational. Visitors can ask questions and get answers grounded in the curator's carefully selected sources.

Think of it as a miniature expert system, built not by prompt engineering but by accumulated human judgment about what matters.

Who's Building Collections

The first Open Collections on Liminary reveal the range of what's possible:

A venture capitalist maintains a collection on breaking into VC careers not generic advice you'd find on Google, but the specific resources, frameworks, and examples he's found genuinely useful over years of fielding questions from aspiring investors.

A nonprofit focused on human agency in technology curates research, policy documents, and articles that inform their work creating a public resource for anyone exploring the same questions.

A security researcher builds a library on AI security threats, synthesizing technical papers, news coverage, and analysis that would take others months to assemble.

None of these people are "content creators" in the traditional sense. They're practitioners who accumulated knowledge through their work and now have a platform to share it.

Building in Public

We're early in this experiment. Liminary launched Published Collections recently, and we're learning alongside our curators what works and what doesn't.

Some early observations:

Collections work best when they're focused. A collection on "AI" is too broad; a collection on "AI applications in pharmaceutical research" serves a real audience.

Context matters as much as content. The best collections include not just sources but the curator's annotations explaining why each piece matters.

Questions reveal gaps. When visitors ask questions the collection can't answer, curators learn what they're missing.

Trust builds slowly. The collections that generate the most engagement are maintained by curators with established credibility in their domains.

An Invitation

If you've spent years learning what actually matters in your field, you already have a collection worth publishing.

You don't need to be a "thought leader" or have a massive following. You just need to have developed genuine expertise in a domain and organized the resources that inform your work.

Liminary handles the technical infrastructure. You provide the judgment and perspective that makes the collection valuable.

The internet has plenty of content. What it needs is more curators willing to say: this matters, and here's why.

Published Collections are live at on.liminary.io. To create your own, start by building a collection in Liminary and simple hit Publish. You can checkout the links used in this article in this published collection: https://on.liminary.io/c/human-curation-renaissanc