You’ve probably seen it by now, even if you didn’t have a word for it. The bland, over-explained, faintly robotic paragraph that says everything and nothing. The email that opens with “I hope this message finds you well” and never recovers. People have started calling this stuff slop: text that’s technically fine and completely forgettable. And here’s the uncomfortable part. A lot of the time, the machine isn’t the problem. We are.
Because a language model doesn’t invent slop out of nowhere. It gives you back a version of what you asked for. Feed it a lazy, shapeless request and it fills the space with the safest, blandest thing it can find. Slop in, slop out. The quality of what comes back is tied, more tightly than most people realise, to the quality of what you put in. And what you put in is language.
Prompting Is Just Talking Well, Under Pressure
There’s a myth that “prompting” is some new technical trick, a set of magic words you memorise. It isn’t. Prompting is just the old skill of saying what you actually mean, made suddenly visible. When you talk to a person, they fill in your gaps. They read your tone, your history, the look on your face. A model has none of that. All it has is your words, exactly as you wrote them. So every vague pronoun, every “make it better” with no sense of what better means, every request that hasn’t decided what it wants, gets exposed.
That’s why the people who get remarkable things out of these tools tend to be the same people who were already good with words. They know the difference between “summarise this” and “pull out the three claims this article is actually making and tell me which one is weakest.” They know that “write me a poem” and “write me eight lines, no rhyme, about missing someone you never met” produce completely different worlds. The skill was never about the AI. It was about knowing your own mind well enough to describe it.
The Fine Gradations Do the Heavy Lifting
What’s fascinating is how small the adjustments are that separate slop from something genuinely useful. It rarely comes down to length or effort. It comes down to precision. A single verb changes everything. “Explain” gives you a lecture; “walk me through” gives you a hand on the shoulder. “Analyse” and “critique” send the model down entirely different roads. Naming your audience does more work than a whole paragraph of instructions: “explain it to a curious ten-year-old” and “explain it to a examiner” are not the same request, and you can feel the difference in the answer.
- The verb sets the direction. Distil, compare, argue, outline — each one is a different job.
- Scope keeps it honest. “Give me three” beats “give me everything” almost every time.
- Register sets the voice. Formal, plain, playful — you choose, or the model chooses for you.
- Naming the reader sharpens all of it. Who is this for? That single answer reshapes the whole reply.
Which Is Really a Story About Learning
If any of this sounds familiar, it should. It’s exactly what we ask students to do when we teach them to write. Choose the right word. Cut the padding. Decide who you’re talking to. Say the thing you mean instead of circling it. The habits that keep slop out of an essay are the same habits that get a good answer out of a model, and it works in both directions: students who practise talking to these tools with care are, without quite noticing, practising the craft of writing itself.
There’s a nice symmetry to how the models get built, too. They’re trained on oceans of text and then patiently corrected, nudged toward answers that are clearer and more useful, one small refinement at a time. That’s not so different from how anyone learns to communicate. You read a lot, you try, you get feedback, you adjust. Fluency isn’t a switch that flips. It’s thousands of tiny corrections that eventually settle into instinct.
So Learn to Ask
The reassuring news is that avoiding slop isn’t a gift some people are born with. It’s a skill, and it’s the same skill whether you’re writing an essay, sending an email, or talking to a machine that answers back. Strong command of English stops being a box to tick on an exam and starts being a lever: the clearer you can make your thinking, the better everything downstream becomes. So the next time you sit down with an AI and get something flat and forgettable, resist the urge to blame the tool. Look at what you asked. Then ask again, better. That habit, more than any clever trick, is what turns slop into something worth reading. Every word really does count.
