What EC Knows
A case study in what language models miss
There is a place called EC.
Not on any map. Not in any encyclopedia. Not in the Wikipedia article about East Cleveland, which mentions Millionaire’s Row, the historic Nela Park, and the median household income of $22,883. EC doesn’t appear there. It doesn’t appear in the Case Western encyclopedia of Cleveland history either, or in the tourism guides, or in the civic databases.
EC is what everybody in the schools calls it. It’s what you call it when you’re from there, or when you’ve worked there long enough to be trusted with the name. It lives in hallways. In the way a kid says where they’re from. In the mouths of teachers who actually showed up.
This week, historian Ada Palmer and cryptographer Bruce Schneier published a piece in The Guardian arguing that large language models have a fundamental blind spot: trained almost entirely on written text, they learned language at its most formal, edited, and archived—and missed the vast majority of how people actually speak. The unscripted conversation. The thing said in a hallway. The name a community gives itself that never quite makes it to the page.
To demonstrate the point, ask an AI what EC means in the context of Cleveland.
I guessed Emergency Center. ChatGPT guessed Emergency Center, European Commission, Executive Committee, and “Each Child—more poetic, but maybe a little on-the-nose.” Both of us working hard. Both of us completely wrong.
The answer was in the schools the whole time.
A few weeks ago, East Cleveland City Schools celebrated the end of years of state academic oversight. They called the celebration We Are EC Day: Three Stars and Rising. Every student got a hoodie. The superintendent said: We scratch. We claw. We grind and we hustle to get what we need.
That event made the local news. In every article, the journalists translated EC back into East Cleveland City Schools within a sentence or two. The community’s own name for itself flickered briefly into the written record and then disappeared again, surrounded by the formal name.
The hoodies are all over EC. They’re nowhere in the data.
What gets lost is not random. The written record has always over-represented the educated, the published, the powerful. Oral traditions, vernacular speech, the names communities give themselves in hallways—these were already underrepresented before AI arrived. The model doesn’t introduce a new bias so much as accelerate an old one.
Multiply EC by every neighborhood, every school, every community that documents itself in celebrations and spoken names rather than text, and the shape of what’s missing starts to become visible.
There is a poem written on April 16, 2026, by someone who had worked in EC:
Not the first “war zone” I have worked in
They are not how you would expect them to be
The smartest children are born in war
Because the community is pulled together
Shared trauma is like a sweatshirt holding you tight
Four lines and one. The one does all the landing.
That poem is not in any training data. It was written this morning by someone standing somewhere knowing something. The only way it enters the record is if someone sends it in.
Which is, perhaps, the beginning of an answer. Not a technical solution—not yet—but a human one. The people who know what EC means have to decide it’s worth telling. The people building these models have to decide it’s worth hearing.
The model only knows what makes it out of the hallway. And the people who could send it often don’t. Or can’t. Or have no reason to think it matters to a machine.
It does.
Nottingham UMC sits in Collinwood, on the north side of Cleveland. The turkey vultures come back to Hinkley every year around the ides of March. EC is what everybody in the schools calls it. It was never missing.
Comments
Post a Comment