The Humanist’s Edge: Why the AI Era Belongs to Those Who Read

There is a group of people who spent the last decade being told their education was worthless. Soft. Impractical. Good for dinner table conversation, perhaps, but not for the real economy. Those people are now, without most of them knowing it, among the best positioned to work with artificial intelligence on the planet. The historians. The philosophers. Those who once argued for three hours over why Rulfo chose one word and not another.

I find this deeply satisfying.

Silicon Valley spent 15 years telling the world to learn how to code. Then it built the most linguistically sophisticated tool in the history of humanity. And that tool responds best precisely to the people Silicon Valley told to retrain.

A New Kind of Interlocutor

Let us set aside the technical side for a moment. Let us think about what it actually feels like to use an LLM (Large Language Model).

You are not searching the web. You are talking to something that has read an enormous quantity of human writing in every language, period, and subject: novels, court rulings, scientific papers, poetry, philosophy, and instruction manuals. It knows how to distinguish between the formal and the colloquial. It catches irony. It recognizes cultural references. And it responds very differently depending on how you formulate your question.

That last point is what most people underestimate. The model does not expect you to configure it like a piece of software. It is reading you. Every word you choose tells it something about who you are, what you know, what register you are operating in, and what kind of response you actually want. It is, in this sense, a profoundly literary instrument.

The quality of what you get depends almost entirely on the quality of what you put in. And that is a problem of language, not of programming.

The Word You Choose Is the Result You Get

A concrete example that has nothing to do with the usual prompting advice.

Suppose you need a text for the relaunch of a brand in the premium segment. You could ask the LLM for something “luxurious.” You would get something generic and faintly glossy. Or you could ask for something, “austere and quietly sure of itself, in the spirit of the late Luis Barragán: clean forms, silence as language, quality that needs no name.”

One sentence. The LLM has absorbed everything that has been written about that aesthetic, its philosophy, its cultural context. It responds accordingly. The difference in quality is not marginal. It is the difference between something that sounds like marketing and something that sounds like a point of view.

This is not a prompting technique. It is vocabulary. And vocabulary is built by reading widely and paying close attention.

The same logic applies to emotional register. There is an enormous difference between asking the LLM for something with “the luminous sobriety of Rulfo’s stories,” something with “the narrative warmth of early García Márquez,” something with “the conceptual precision of a Borges essay,” or something with “the urban irony of Monsiváis.” These are not tone instructions. They are cultural references the model knows in depth, and which convey in a single sentence what three paragraphs of explanation would not manage.

The Cultural Dimension

Add cultural knowledge to linguistic precision and the advantage multiplies. This is where my own background becomes especially relevant.

Twenty years working across Mexico, Latin America, Europe, the Middle East, and Asia taught me one consistent lesson: the people who handled intercultural situations best were not the most technical. They were the most attuned to context. They understood that words do not travel clean across cultures. That the same concept, translated literally, can mean something entirely different on the other side.

A real example. The English word “accountability” is culturally loaded in a very specific, Anglophone way. It has no clean equivalent in Spanish, and everyone in the professional world knows it. “Responsabilidad” is too passive. “Rendición de cuentas” is too bureaucratic and carries connotations of auditing and punishment rather than genuine ownership. Whoever understands this gap can orient the LLM in a very different way:

“The concept I am working with is closer to proactive ownership than to the Anglophone notion of accountability. In Spanish, the closest approximation would be something like hacerse cargo. Build the framework around that cultural intuition, not around the English word.”

The LLM knows exactly what “hacerse cargo” implies. It adjusts the entire approach accordingly. Someone working only in English would not think to flag that gap. They would receive something that uses “accountability” throughout the document, a text to be read by people for whom the underlying concept is culturally foreign.

The model does not merely respond to what you say. It responds to what you know.

Tell the Model Who You Are

This leads to something I believe most people overlook entirely.

Because the model reads you through the language you use, the most efficient thing you can do before any serious conversation is to give it a precise picture of who it is talking to. Not a professional biography. A cultural and linguistic portrait of yourself.

Rather than writing “I prefer clear communication,” a humanist might write:

“I write directly and without adornment, in the spirit of the best essays of Octavio Paz: clarity as a gesture of respect toward the reader, irony as a tool and not as ornament. My references are closer to Borges than to consulting presentations.”

The LLM knows Paz, Borges, and the Latin American essay tradition in depth. That one sentence carries more information than three paragraphs of tone instructions, and produces a qualitatively different response.

The humanist already understands this instinctively. Context is not the background. It is the entire conversation.

Beyond Writing: The Case of Coding

The obvious objection: This is all well and good for writing, but coding is purely technical.

It is not, and the best developers working with AI will say the same thing. The hard part of building something with an LLM is not knowing the programming language. It is being able to specify what you want with enough precision for the model to deliver it. The programmer who can describe the behavior of an edge case precisely, in plain language, gets better code than the one who cannot. That is a writing skill.

More than that: A programmer who can say “this function should fail noisily, not silently” is making a philosophical point about error design. The model understands it immediately. Someone without that vocabulary describes the same thing in twenty words and still does not quite say it.

The bottleneck in coding with AI assistance is almost never technical knowledge. It is almost always clarity of thought expressed in language.

What All of This Means

Over the last decade, the standard response to AI in education has been to cut the humanities and redirect resources toward Science, Technology, Engineering, and Mathematics, the cluster of disciplines known as STEM. Understandable. Also, in the context of AI, exactly backwards.

The people and organizations that will extract the most from these tools over the next decade are those with the greatest command of language, the broadest cultural literacy, and the habit of reading attentively. Not as a complement to technical skills. Often, as the primary driver of quality outcomes.

Technical fluency without linguistic depth produces competent operators. Linguistic depth combined with technical fluency produces people who can unlock what these systems are truly capable of.

There is a satisfying irony in all of this. The subjects most aggressively cut, the most systematically dismissed as impractical, turn out to be structural advantages in the AI era. The Latin students. The philosophers. The historians. Those who read Paz, Rulfo, Borges with attention and argued over why a single word mattered.

They have been practicing the right skills all along. They just did not know what they were practicing for.