A step-by-step guide to turning a ChatGPT draft into an essay that reads like you wrote it — without breaking the argument or your citations.

You asked ChatGPT for a first draft. It's coherent, it hits every point of the rubric, and it reads like every other ChatGPT essay your professor has seen this semester. Here is the exact workflow to turn that draft into something that sounds like you — and passes a reasonable AI check — without losing the argument.

Step 1: Own the argument before you touch the prose

Read the draft once, out loud, without editing. Then close it and write the thesis in one sentence in your own words. If you cannot, the model wrote an essay you cannot defend, and rewriting sentence-level is a waste of time. Regenerate with a tighter prompt or write the thesis by hand and let the model fill in around it.

Step 2: Break the paragraph rhythm

The single strongest AI tell is uniform sentence length. ChatGPT tends to write in 19-to-24-word units back to back. Real students mix a 30-word sentence with a five-word sentence in the same paragraph. Go through and split at least one long sentence per paragraph into a fragment. Merge two short ones with a comma. Vary the openers — never open three sentences in a row with the same part of speech.

Step 3: Kill the connector words

"Furthermore", "moreover", "additionally", "in conclusion" — these are the vocabulary of a five-paragraph essay generator, not a college student. Delete them. The sentence that follows almost always works without the connector, and the paragraph reads faster.

Step 4: Replace one abstract noun per paragraph

AI writing is abstract-noun-heavy: "implementation", "consideration", "utilization", "understanding". Swap one per paragraph for the concrete verb it hides. "The implementation of the policy" becomes "when the school put the policy in place". Your prose immediately drops a register and sounds human.

Step 5: Add one specific detail per section

Models generalize by default. If your essay is about the French Revolution, they'll write "many peasants" instead of "the villagers of Ruffec, who paid a bread tax of…". Even one named person, place, or number per section makes the essay read as if you did the reading.

Step 6: Fix the em-dash problem

ChatGPT loves em-dashes — sometimes three per paragraph. Real academic writing uses them sparingly. Replace two out of three with a comma or a period. Bypassify does this for you automatically, but doing one pass by hand teaches you where they cluster.

Step 7: Protect your citations

The riskiest part of any AI draft is the citations. Models hallucinate page numbers, invent authors, and swap edition years. Every citation in your final essay has to be one you can open. If you use a humanizer, use one that preserves quotes and citations verbatim (that is the whole point of our preserve-quotes toggle) — otherwise you might change "(Smith, 2019, p. 42)" into "(Smith, 2019, p. 45)" and ruin an otherwise good essay.

Step 8: Read it out loud one more time

The last pass is auditory. If a sentence doesn't sound like something you'd say in class, rewrite it. If a paragraph sounds like a Wikipedia summary, cut the first sentence — it's probably a topic sentence the model added and you don't need. What survives will read like you.

The shortcut

You can do all eight steps by hand — it takes about 40 minutes for a 1200-word essay. Or you can paste the draft into Bypassify, which runs steps 2 through 6 automatically while protecting citations, then do steps 1, 5 and 8 by hand in ten minutes. Either way, don't skip the last read-through. That's what makes it yours.

How to Humanize a ChatGPT Essay Without Losing Your Grade