The AI in the room

On losing a friend to ChatGPT

An old friend asked me to review her statement of purpose (SOP) for a PhD program. She sent me the Google Doc link and said “be brutal.” So I read it.

It did not sound like her. It sounded like something ChatGPT would write if you gave it a prompt that said “write a compelling SOP for a PhD program.” Every paragraph was perfectly structured, every transition was smooth, every sentence sounded important. It was littered with em dashes, way more than anyone naturally uses. The contractions were everywhere, “I’m” and “it’s” all over a formal academic document. And there was something else I could not quite name. It was not that it was too polished. I just cannot name it. It did not feel like an SOP, it felt like an essay that was performing the idea of an SOP.

So I told her. I said it reads like AI slop.

She did not take it well. First, she misread “AI slop” as “AI SOP” and took that to ChatGPT to understand what I meant. I corrected her. AI slop is a term people use when something looks and feels AI generated. Then she went back to ChatGPT a second time, this time to ask if “AI slop” was valid feedback. ChatGPT told her it was “rude and lazy.” She came back with that energy. Then she blocked me on LinkedIn, Twitter, and Instagram.

Over an SOP review she asked me for.

Let me say that again. She asked a human being for feedback, did not like what the human being said, went to an AI TWICE to make sense of it, and then used the AI’s opinion to dismiss the human being. The same AI that probably helped write the SOP in the first place.

There is a saying I once heard that stuck with me. If I need help and I come to you for it, and you refuse, I will not be angry. You know why? Because if it were easy, I would have done it myself. The reason I asked is because it was hard. So whether you help me or not, I will figure it out.

But here is what I think people get wrong. That same grace should apply when someone actually helps you and you do not like what they say. She came to me because she needed a second pair of eyes, she could not see the problem herself. That is why she asked. So when I told her what I saw, that should not have been the end of the friendship. That should have been the beginning of a better draft.

I have been thinking about this since it happened. Something is shifting in how we relate to each other and I do not like it.

We have started outsourcing our judgment to AI. Someone says something that stings, and instead of sitting with it, we run to ChatGPT for a second opinion. The problem is that ChatGPT will almost always tell you what you want to hear. It does not know your friend stayed up until 1 AM reading your document. It does not know that your friend has a PhD and has been through this exact process. It does not know that your friend followed up with specific feedback even after you started pulling away. It just sees text and optimizes for making you feel better.

AI is an amplifier. If you give it a well thought out draft, it will make it sharper. If you give it garbage, it will make the garbage sound eloquent. And that is exactly what happened with her SOP. The ideas underneath were real, her experience was real, but she let AI smooth out every rough edge until there was nothing left that sounded like a human being wrote it.

And then when a human being pointed that out, she chose the machine’s opinion over the friend’s.

I think about this beyond just writing. We are in an era where AI can do a lot, but it cannot replace the people in your life who will tell you the truth. Those people are rare. They are more valuable than any tool. And if your first instinct when someone gives you hard feedback is to go find a bot that will disagree with them, you are going to surround yourself with comfort and call it support.

Some people want help. Other people want validation. It is important to know the difference before you invest your time. This is why adult friendships are tough. When you are younger, friendship survives on proximity and good vibes. But as you get older, it gets tested by honesty. And some people will choose comfort over a friend who tells them the truth.

Maybe she was never really a friend after all. You don’t discard your friend that way. You don’t block someone across all platforms over feedback you asked for, run to ChatGPT twice to tell you it was “rude and lazy,” unless they were never really a friend to begin with. Or maybe she was, and she just was not ready. Either way, I wish her well. I hope she rewrites that SOP in her own voice. It would be a good one if she did.

Originally published as a Daily Insight on tobilobaadejumo.com on February 10, 2026.