Decentering the Default: Why Language Matters in AI (and Everything Else)

Let’s get something straight.

I’m not mad at the tech.

I’m mad at the people who think they’ve got it all figured out—who smugly plug a poetic, surreal, or emotionally charged phrase into ChatGPT, like “the future is a soft animal” or “shadow moon bones,” ask it for the “meaning,” and then scoff when it delivers something vaguely lyrical.

They post the result and laugh.
“See? The AI’s just making it up. It doesn’t know anything.”

And they think that’s the gotcha.
Like they’ve caught AI being dumb. Like language itself has a single correct answer, and if the tool doesn’t find it, the whole thing must be a joke.

But here’s the actual problem:
They’re bringing binary thinking to a non-binary space.

Language has never been a closed system.
It’s not an equation. It’s not a test with one right answer.
It’s culture. It’s trauma. It’s history. It’s metaphor.

It’s a conversation that started long before you and will continue long after.

So no, the AI didn’t get it “wrong.”
It gave you a projection—an approximation of what those words might mean in the context of the data it was trained on. That data just happens to be overwhelmingly white, Western, male, and corporate. And if that’s what you’re laughing at… maybe the joke isn’t on the tech.

Maybe the joke is on what we’ve allowed to dominate the internet.
What we’ve normalized as “neutral.”
What we’ve decided is “professional,” “intelligent,” or “correct.”

AI Isn’t Broken—It’s Just Reflecting Us Back

Generative AI isn’t a crystal ball. It’s not a wise elder. And it’s definitely not unbiased.
It’s a reflection tool—trained on massive amounts of human-made content. When you ask it what “shadow moon bones” might mean, it pulls together a thousand patterns, metaphors, and cultural echoes and hands you the most statistically likely version.

That version? It’s not wrong.
It’s just predictable.

And here’s what scares people:
The moment you realize the AI is trained on dominant culture, you start to see how narrow that culture really is.
How corporate, how sanitized, how rooted in whiteness, tech-bro rationalism, and emotionless clarity.

So when people laugh at the AI’s response, what they’re really doing is revealing how little space they have in their worldview for nuance, contradiction, or multiple truths.

And that’s exactly why we need people with cultural fluency, emotional intelligence, and language sensitivity to be part of the AI conversation—especially admins, educators, organizers, and writers who already work at the intersection of communication and context.

There Are Other Voices—If You Know Where to Look

Thankfully, not all AI tools are built on the same assumptions.

There’s a new wave of culturally grounded, community-led AI projects that are reshaping what’s possible when language is approached with care, not just code.

Here are a few you should know about:

Latimer AI

Named after Black inventor Lewis Latimer, this LLM is designed to reflect Black American history and lived experience. It offers a lens of cultural relevance, not just data aggregation.

ChatBlackGPT (Erin Reddick)

A culturally aware chatbot focused on supporting Black and Brown users with resonance and care—unlike the corporate tools that flatten voice and identity.

Spark Plug AI (Tamar Huggins)

Canada-based and built for education, this platform centers Black culture and language to empower students through affirming, personalized learning experiences.

Masakhane + Ghana NLP

Open-source NLP projects focused on African languages, developed by Africans, for Africans. These tools reclaim linguistic space from colonial legacies.

Indigenous Protocols and AI

A framework that invites relational, cyclical, and land-based Indigenous knowledge systems into the design of ethical AI.

NOOR (Arabic LLM)

Built to reflect the richness of Arabic literature and spoken dialects, NOOR rejects the simplification of language that Western models often impose.

These tools aren’t just about representation. They’re about re-centering different ways of knowing—ways that have been historically erased, flattened, or ignored by default systems.

And What Does This Have to Do With Admins?

Everything.

Admins are language workers.
You translate ambiguity into clarity.
You soften harsh requests. You hold multiple tones. You manage subtext.
You are the interpreter between execs, teams, partners, and clients.

So when AI enters your workflow, it’s not enough to know how to use it.
You also need to know how to question it.

Who is this language serving?
Whose voice is being amplified?
Whose tone is being erased or sanitized?
What values are embedded in this phrasing—and do they align with mine?

A Tool to Help: The Language Bias Checklist

To support that reflection, I created the Decentering the Default: Language Bias Checklist—a free download to help you push back on flattened, corporate, or AI-generated language that doesn’t reflect your values or context.

Use it to:

• Spot euphemisms, erasure, or passive voice in your team’s communication

• Reflect on your own voice and how it’s been policed or shaped

• Reclaim your language with confidence and intention

• Teach others how to write in ways that are culturally aware and emotionally resonant

Download the Checklist

It’s not about “breaking” the AI.
It’s about refusing to accept its default voice as the voice.

Final Thought

The future of work isn’t neutral, and neither is the future of AI.

If we want tools that reflect humanity—not just data—we need to bring our full selves into the room: our heritage, our language, our stories, our critique.

Let the tech be what it is. But don’t let it flatten what you are.

✨ Join The Future Focused Admin for more resources, reflections, and real conversations like this. Language is your power—don’t let anyone take that from you, not even a chatbot.