Teacher Takes Notes

Here’s a fact that surprises most people: several of the people who built Anthropic — the company behind the AI you might be using right now — studied literature, not computer science. Daniela Amodei, Anthropic’s president and co-founder, has a degree in English Literature. So does Jack Clark, another co-founder, who worked as a journalist before helping build one of the world’s leading AI labs. (Anthropic’s CEO, Dario Amodei, does have a technical background — a PhD in biophysics — so it’s not that literature replaced science at the company. It’s that both turned out to matter.)

That’s not a coincidence, and it’s not just an interesting trivia fact. It points to something real about how AI models actually work — and why the skills you’re building in Language and Literature class are becoming some of the most useful skills in tech.

Talking to an AI Is Not Coding. It’s Communicating.

A well-known line in the AI world, from OpenAI co-founder Andrej Karpathy:

English has become the hottest new programming language.

He wasn’t joking. For decades, getting a computer to do something meant learning a rigid, symbolic language — Python, Java, C++ — with strict rules and zero tolerance for ambiguity. Now, the main way people instruct AI models is by writing sentences. Plain, ordinary language.
That sounds like it should make things easier. In one sense it does — anyone can type a sentence. But in another sense, it raises the bar, because ordinary language is full of ambiguity, tone, implication, and assumption — exactly the things you spend English class learning to control.

A Prompt Is a Speech Act
Linguists have a term for what a prompt actually is: a directive speech act, the same category of language as asking someone to pass the salt, or asking a friend to open a window. Directive speech acts aren’t just information; they’re an attempt to get someone (or something) to do something, and how you phrase them shapes whether they land.

Think about the difference between:
• “Can you maybe help with this if you get a chance?”
• “Please summarize this article in three bullet points by the end of the message.”

The first is polite but vague — a human listener would guess at what you actually want, and an AI model will do the same, often incorrectly. The second gives a clear task, a clear format, and a clear boundary. Researchers studying prompt design have found this same pattern holds when working with AI: models respond best to instructions that specify purpose, audience, format, and constraints — the exact elements you already analyze when you study a text’s rhetorical situation.
Register, Specificity, and Structure — Sound Familiar?
If you’ve been following the “informal vs. formal register” ideas from our last post, you’ll recognize these same moves showing up in prompt writing:
Vague, informal request:
“write something about climate change”
Precise, purposeful prompt:
“Write a 150-word summary of the main causes of climate change, aimed at a general audience with no scientific background. Use plain language and avoid jargon.”
The second version does exactly what a strong formal paragraph does: it names its audience, defines its scope, and controls its register — “plain language, no jargon” is itself a register instruction. Prompt engineers call this “specificity.” You’d call it knowing your purpose and audience — Criterion C, if you want to put an MYP label on it.
There’s also a structural echo of essay-writing in how AI models respond to reasoning tasks. One widely used technique, called chain-of-thought prompting, asks the model to work through a problem step by step rather than jumping to an answer — essentially, asking it to show its reasoning the way you’d be asked to in a structured paragraph, with claims built on evidence built on explanation.

Why Linguists — and Language Students — Have an Edge
Researchers who study language and AI together have pointed out that prompt engineering draws heavily on pragmatics, the branch of linguistics concerned with how context shapes meaning — the same skill you use when you infer a character’s real intention behind what they say, or when you adjust your tone for a teacher versus a friend. One writer on the topic put it simply: prompting isn’t really about talking to machines, it’s about translating human meaning into a form a machine can act on. That translation work — finding the clearest, most precise way to say what you mean — is a literacy skill before it’s a technical one.

There’s a real limitation worth knowing too: this fluency doesn’t transfer equally across languages. Studies on multilingual prompting have found that some languages require more words (and more computing power) to express the same idea, and prompting techniques proven in English don’t always work the same way in a language with different word order or different ways of marking formality — a reminder that language itself, not just wording, shapes how well an AI understands you.

Try It Yourself
Take a task you might actually give an AI tool — say, asking for help planning a study schedule. Compare:
Informal / vague: “help me study for my test”
Precise / well-formed: “I have a Language and Literature test on Friday covering the novel Ghost by Jason Reynolds — themes, characters, and one persuasive-writing skill. Create a 3-day study plan, one task per day, each taking no more than 30 minutes.”
Notice what changed: audience and purpose became explicit, scope got defined, and format got specified — precision, not decoration, is what makes the second version work.
Advanced Moves: Using Your Language Skills to Collaborate With AI
If you’re in MYP Language and Literature, you already have tools most prompt writers don’t. Here’s how to use them deliberately — and how to do it in a way that strengthens your own writing rather than replacing it.

1. Prompt the way you annotate a text.
   When you close-read a poem or a passage, you don’t just react — you name what you’re noticing and why it matters (imagery, structure, tone). Apply the same habit to AI feedback. Instead of asking an AI to “check my essay,” ask it to evaluate a specific element: “Look only at my topic sentences — are they clearly linked to my thesis?” or “Identify where my register shifts from formal to conversational.” A narrow, well-defined question gets a far more useful answer than a broad one — the same way a focused thesis produces a stronger essay than a vague one.
2. Borrow criteria language on purpose.
   You already have a shared vocabulary for what “good” looks like in this subject — the MYP criteria strands (organization, register, literary devices, textual evidence). Use those exact terms in your prompts: “Using MYP Criterion D language, tell me where my word choice feels too informal for a literary essay.” Naming the standard you’re being judged against produces feedback that’s actually usable, instead of generic praise.
3. Draft in your own voice first, revise with AI second.
   This is a direct extension of the “formality pass” idea from before: write your first draft fast, in your own thinking voice, to get your real ideas down. Then bring in AI — not to write for you, but to interrogate what you’ve already written: ask it to flag unclear sentences, point out where an idea needs more evidence, or ask you a question you haven’t answered yet. Used this way, AI functions like a very fast peer-reviewer, not a ghostwriter — and your ideas, argument, and voice stay yours, which matters both for your growth as a writer and for academic integrity.
4. Iterate the way you redraft — treat the first answer as a draft, not a verdict.
   Skilled writers don’t accept their first sentence as final, and skilled prompt writers don’t accept an AI’s first response as final either. If an answer feels off, don’t start over — refine, the way you’d revise a paragraph: “That’s too generic — give me an example specific to the novel we’re studying,” or “Make this more concise, one sentence instead of three.” Precise revision requests come from the same skill as precise self-editing.
5. Interrogate tone and audience, not just content.
   You already ask “who is this written for, and why?” when you analyze a text. Ask AI the same question about its own output: “Who does this response sound like it’s written for? Is that the right audience for a Grade 9 reader?” This catches a common problem — AI responses that are technically correct but pitched at the wrong register or reading level for the task you actually need.
6. Stay the editor-in-chief.
   The most advanced prompting skill isn’t a phrasing trick — it’s judgment. A strong reader can tell when a text is overstating its evidence, contradicting itself, or making a claim it hasn’t earned. Bring that same critical eye to AI output: fact-check claims, question generalizations, and notice when something sounds authoritative but isn’t actually well-supported. The language skills that make you a sharp reader of novels make you a sharp reader of AI text too — and that judgment is exactly what stays yours, no matter how good the tools get.
   The Takeaway
   You don’t need to learn to code to work well with AI. You need to do what you already do in this class: read a situation, identify your audience and purpose, choose your words with precision, and structure your thinking clearly. The people who built the AI you’re using bet on that — some of them, quite literally, with an English degree.