The Vocabulary Gap
Why the Best AI Users Sound Like Philosophers — And What That Tells Us About Where We Actually Are
There’s something I’ve noticed about the people who are genuinely good at using AI — not the people who post screenshots of ChatGPT writing their emails, but the people who’ve internalized inference-based reasoning as a cognitive tool. The ones who’ve been in the trenches long enough that it’s changed the way they think.
They all sound like philosophers.
I know because I do it. I catch myself in conversations with other practitioners, and we’re three levels deep into abstraction before either of us realizes we’ve left the building. We’re talking about “shaping output spaces” and “co-cognition” and “the texture of a reasoning chain.” We’re using metaphors stacked on metaphors. We’re gesturing at something real — something we’ve both experienced — but the words keep sliding off it like water off glass.
I used to think this was just what happens when technical people get excited. But I don’t think that’s what’s going on. I think something more fundamental is happening, and I think it matters.
We Don’t Have the Words Yet
Here’s my theory: the reason proficient AI users default to abstract, philosophical language when discussing AI is that we literally don’t have the vocabulary for what we’re experiencing.
Modern inference AI — the kind you actually interact with, the kind that reasons — is roughly three years old in any meaningful public sense. Three years. That’s nothing. That’s not enough time for a culture to develop the language infrastructure needed to describe a fundamentally new category of human experience.
Think about what happens when you sit down with a capable model and do real work. Not asking it to summarize an article. Real work — the kind where you’re iterating on a complex problem, where the model pushes back on your assumptions, where you adjust your prompting mid-stream because you can feel the reasoning drifting, where the output surprises you in a way that reshapes your own thinking.
What is that? What do you call that experience?
It’s not programming. It’s not searching. It’s not having a conversation. It’s not delegation. It’s not collaboration in the way we use that word between humans. It is something genuinely new — a form of directed cognitive partnership that has no name, no clean taxonomy, no agreed-upon set of descriptors.
So what do we do? We reach for the closest thing we have. And the closest thing we have is philosophy.
This Has Happened Before
This isn’t a new problem. It’s a very old problem wearing new clothes.
When the automobile first appeared, people called it a “horseless carriage.” Not because that was a good description — it tells you almost nothing about what a car actually is or does — but because the only reference frame available was the thing it was replacing. The entire vocabulary of personal transportation was built around horses. So the car got described in terms of horse-absence rather than car-presence.
Photography did the same thing. Early photographers borrowed the entire language of painting — composition, exposure, portraiture — because there was no language of photography yet. It took decades for photography to develop its own critical vocabulary, its own way of talking about what made a photograph good or bad on its own terms rather than as a derivative of painting.
The internet went through an even more dramatic version of this. We “surfed” the web. We visited “pages.” We went to “sites.” We put things in “folders.” Every single one of those terms was a metaphor borrowed from the physical world because the digital world didn’t have language yet. Some of those metaphors stuck. Some of them actively limited how we thought about what was possible. It took years before we had words like “streaming,” “cloud,” “viral” — terms that described digital-native phenomena on their own terms.
AI is in the horseless carriage phase right now. And the people who are furthest ahead in actually using it are the ones who feel the vocabulary gap most acutely.
The Double Gap
But here’s where it gets more interesting. The vocabulary gap is the primary driver of the philosophical drift, but it’s not the only one. There’s a second gap underneath it that makes the problem worse: the experiential gap.
When a proficient AI user tries to explain what they do to someone who hasn’t spent real time with these systems, they’re not just missing words. They’re missing shared experience. It’s like trying to describe a flow state to someone who’s never been in one. You can use every word in the English language and still not bridge the gap, because the listener doesn’t have the felt sense that gives those words meaning.
So the speaker goes abstract. Not because they’re being pretentious. Not because they’re trying to gatekeep. Because abstraction is the only level at which communication is even possible when the concrete referents aren’t shared.
This is why two proficient users can sit in a room, speak in what sounds like pure philosophy to an outsider, and walk away feeling like they had one of the most precise, productive conversations of their week. They’re not being vague with each other. They’re being maximally efficient — using shared experiential anchors to communicate at a bandwidth that plain language can’t support yet.
What We’re Actually Missing
Let me get specific about what vocabulary we don’t have. These are real phenomena that practitioners experience daily but have no clean words for:
The thing where you can feel a model’s reasoning quality shift mid-response. There’s a texture to it. Experienced users detect it instinctively — a subtle flattening, a loss of coherence, a drift toward generic language. We call it things like “the model lost the thread” or “it’s pattern-matching now instead of reasoning.” But those are descriptions, not terms. There’s no word for this.
The skill of structuring a prompt so that the model’s inference path is shaped before it begins generating. This isn’t “prompt engineering” — that term has been diluted to meaninglessness. This is something closer to setting up a cognitive scaffold that the model builds within. It’s architectural. It’s strategic. It has no name.
The moment when a model’s output genuinely changes your thinking about a problem. Not because it told you a fact you didn’t know, but because the structure of its reasoning revealed an angle you hadn’t considered. This is qualitatively different from reading a book or talking to an expert. It has no name.
The learned intuition for which tasks are inference-shaped and which aren’t. Proficient users develop a sense — almost unconscious — for when AI will be transformative versus when it will be a waste of time. This is a real skill. It’s arguably the most important skill in professional AI usage. It has no name.
Every one of these is real. Every one of these is experienced by thousands of practitioners daily. And every one of them forces the practitioner into philosophical language because there is simply no precise, commonly understood term available.
Why This Matters More Than You Think
This isn’t just a linguistic curiosity. The vocabulary gap has real consequences.
It slows adoption. When the best users of a technology can’t articulate what they’re doing in concrete terms, it becomes incredibly difficult to train others. AI literacy programs — and I run one, so I feel this in my bones — constantly run into the problem of translating practitioner intuition into teachable frameworks. The absence of shared vocabulary makes every training session harder than it needs to be.
It distorts the public conversation. When practitioners default to philosophical abstraction, they inadvertently reinforce the narrative that AI is mysterious, unknowable, almost mystical. This feeds both the utopian hype and the existential panic. Neither is useful. Both are partially products of a vocabulary failure.
It creates a false hierarchy. If the only way to talk about AI proficiency is through abstract, philosophical language, then the conversation becomes inaccessible to people who think concretely. This isn’t an intelligence divide — it’s a communication failure. There are brilliant, concrete thinkers who could become exceptional AI users but bounce off the discourse because it sounds like a graduate seminar in phenomenology.
It advantages whoever fills the gap. And this is the strategic point. The people who develop the vocabulary — who coin the terms that stick, who create the frameworks that others adopt — will have outsized influence on how the entire field develops. Language doesn’t just describe reality. It shapes what we can think. The vocabulary we eventually build for AI will determine, in part, what AI becomes.
The Path Forward
I don’t think the solution is to stop being philosophical. The abstractions are doing real work right now — they’re the best tools we have for communicating about genuinely novel experiences. Forcing premature concreteness would be worse than tolerating productive abstraction.
But I do think we should be conscious of what’s happening. When you find yourself reaching for metaphor after metaphor in an AI conversation, recognize that for what it is: evidence that you’ve encountered something real that language hasn’t caught up to yet. That’s not a failure of your thinking. It’s a feature of being early.
And for those building in this space — the researchers, the product people, the educators, the founders — there’s an opportunity here that most people are sleeping on. The vocabulary is going to get built. It always does. The automobile got its own language. Photography got its own language. The internet got its own language. AI will get its own language.
The question is whether it gets built deliberately, by people who understand what they’re naming, or whether it accretes accidentally from marketing decks and Twitter threads.
I know which one I’d bet on producing better outcomes.
Bill Faruki is the Founder & CEO of MindHYVE™ and Chair of the California Institute of AI. He has been building agentic AI systems since 2022.