Columbia Supplemental Essays: How to Edit an AI Draft Without Losing Your Voice
Columbia’s supplement list rewards specific intellectual curiosity—not generic NYC love letters. Edit AI drafts for checkable detail. No logos, no endorsement, no invented reading lists.
Columbia’s supplemental stack mixes list prompts, short answers, and fit questions. ChatGPT treats them like one essay: polished, interchangeable, and weirdly enthusiastic about “the vibrant city of New York” without a block, a book, or a lab you actually used.
This is an editing guide for AI-assisted drafts. Human Writes is not affiliated with Columbia University. No logos. No doorway “Columbia AI humanizer” framing.
Hub: personal statement guide, Harvard supplements, MIT short answers, ethical AI checklist.
What Columbia-style prompts punish
| AI habit | Stronger move |
|---|---|
| Generic NYC praise | One place tied to your habit or question |
| Reading list from Goodreads top 100 | Books you can discuss in an interview |
| “Core Curriculum” name-drop without a text | A line you argued about in class |
| Even list rhythm | Different energy per item |
| Resume in prose | One thread; leave the rest to activities |
International applicants: do not let the model Americanize your story. Specificity beats idioms. See international student research guide for voice false-positive context—not an excuse to skip substance.
List prompts: one line, one proof
For “books / publications / art” style lists:
- Item you actually consumed
- One clause on what it changed in your thinking—not a plot summary
- Cut anything you would fail to discuss for five minutes
| Weak | Stronger |
|---|---|
| 1984 – important dystopia | 1984 – made me notice how our school newspaper softens admin quotes |
| The New York Times | NYT Op-Docs, Feb piece on transit—question I reused in math modeling |
Short fit prompts
When a prompt asks why Columbia or why your field:
- Open a real syllabus or catalog page you read this month.
- Name one faculty paper or institute you understand—not twelve.
- Say what you would do, not what you admire.
If the fact is not on a page you can point to, delete it.
Before and after
AI list line:
A Brief History of Time – this inspiring work taught me to appreciate the wonders of the universe and the power of scientific inquiry.
After your read:
A Brief History of Time – the chapter on expanding space finally made sense of what our physics teacher demo’d with the balloon; I kept asking why the analogy breaks near c.
Workflow
- Separate each supplement into its own doc.
- Highlight claims. Delete unverifiable ones.
- Write the first line of each response yourself.
- Match word caps; cut adjectives first.
- One Human Writes pass per essay if allowed—not all prompts at once.
- Read aloud: does every list item survive “tell me more?”
What not to do
- Invent summer programs or internships.
- Paste the same “why NYC” paragraph into every prompt.
- Use Columbia logos or imply endorsement.
- Humanize five times until your specific details disappear.
Bottom line
Columbia supplements reward intellectual specificity, not skyline poetry. Lock true books, places, and questions, then one careful polish pass.
Paste stiff list lines on Human Writes one prompt at a time.