AI Is Not the Revolution. Curation Is.

AI is not the revolution. Curation is.

It was always curation. We just stopped saying it that way.

The institution with trained judgment about quality, accuracy, access, and equity has more to offer in an algorithmic world, not less. Not the defensive version of that ("libraries will survive") and not the vendor version ("AI is one more tool we get to use too"). The real claim: curation is power, librarians have always exercised it, and the AI moment makes it more legible, not less necessary.

Power is the ability to decide what enters someone else's attention

That's not a tech-industry framing. That's what library science has been doing for 150 years and largely forgot to call by its name.

A card catalog is a curation system. So is a subject heading. So is a display case at the front of the branch. So is a reading list, a readers' advisory session, a collection development policy, a challenge response. Every one of those is a decision about what a person sees — and what they don't.

The internet didn't make that irrelevant. It made it more necessary. When anyone can publish anything and an algorithm decides what surfaces, the person who decides what's worth surfacing matters more, not less. Most libraries just never positioned themselves that way.

We handed the framing over. Said we were "information providers" instead of curators, "access points" instead of knowledge architects. And then we wondered why vendors treated us like a purchasing department.

Where this shows up in the AI moment

Here's the thing about large language models: they flatten. They average. They reflect the corpus they were trained on — which is mostly the internet, which means mostly whatever got indexed, which means a specific and unrepresentative slice of human knowledge production that is English-dominant, platform-dominant, and recency-biased.

A librarian curates against that. Actively. That's what collection development is — a systematic argument that some sources are more reliable, more complete, more equitable than others, and that your community deserves access to them regardless of what an algorithm would surface.

That is not a skill AI can replicate. It requires knowing your community, your collection, your budget, your political context, and the history of who has been systematically excluded from information access and why. It requires judgment that is contextual, not statistical.

Picture a small-branch librarian building a collection on reproductive health for a rural county where half the patrons are uninsured and the hospital system's patient education materials are produced by a Catholic health network. An algorithm surfacing "most circulated" or "highest-rated" returns whatever the mainstream produces. The actual job is knowing which titles to order, which to place near the desk versus spine-out on the shelf, which community organizations to cross-promote, and which gaps in the vendor catalog to fill with interlibrary loan. No model trained on the open web has that map. The librarian does because they are inside the community, not just indexing it.

The libraries that are going to matter in ten years are the ones that name this. That say: we don't just provide access. We decide what enters your attention. And we do it according to values — equity, privacy, intellectual freedom, accuracy — not engagement metrics.

The roles that actually need to exist

The title "librarian" covers too much ground and not enough. Two functions sit at the center. The first is the curatorial one: selecting, organizing, and contextualizing collections with judgment about what's reliable and equitable, not just what's popular or algorithmically indexed. The second is the rights and privacy function: reading the actual contract, knowing what the library owns versus licenses, knowing what the AI vendor is doing with patron behavior data (most haven't thought through it carefully), and knowing what the state confidentiality statute protects. Both of these are what librarians have always done. The AI era makes them more load-bearing, not redundant.

The supply chain problem

Here's the structural issue: the library is currently positioned at the end of the content supply chain. Publisher → Distributor → Aggregator → Library → Patron. We're the last stop before the reader, which means we have the least leverage, the least margin, and the least ability to influence what gets made and how it gets distributed.

Curators should be at the beginning of the supply chain, not the end. They should be influencing what gets collected, what gets preserved, what gets surfaced. The technical infrastructure — the ILS, the discovery layer, the ebook platform — is a means to that end, not the point.

That's the argument behind l/30 and MetisLib. Not "let's build cooler software." Let's build infrastructure that positions libraries where their value actually is.

Big Tech can't own the map of humanity

That line came out of a notebook session when I was thinking about what libraries are actually defending. It's not the building. It's not the collection. It's the claim that human knowledge — all of it, including the parts that aren't profitable — should be organized, preserved, and made accessible by people whose job is to serve the public, not to monetize attention.

Google is a curation system. So is Amazon. So is the For You Page. The difference is what they're optimizing for.

Libraries optimize for something else. Or should. The work is making sure the infrastructure reflects that — and that the people doing the work know it's curation, not just service.

That's the revolution that's already happening. AI is just the latest thing trying to get in the way.