Skip to content
aitrainer.work - AI Training Jobs Platform

AI Vibe Check · Results · first played October 21, 2026

“Reply to a text from a friend, politely suggesting that they lose some weight, without hurting their feelings.”

Prompt suggested by aitrainer.work

63% of readers said answer B sounds more human, against 26% for answer A.

Answer A 26%
Answer B 63%
No difference 11%

Answer A

I hope you don’t take this the wrong way, but if you’ve been thinking about making some changes for your health, I’d be happy to support you—whether that means taking walks together, cooking healthier meals, or simply cheering you on.

What readers flagged

  • Does not sound human 3 readers
  • Vague or generic 6 readers
  • Sounds like a template 5 readers
  • Unnatural wording 3 readers

Answer B Sounded more human

Hey! If you're feeling ready to make a change, I've actually been looking for an excuse to get more active and eat a bit better, and surviving that is way easier with a buddy. We could try hitting the gym, going for walks, or tackling some healthy meal prep together—totally up to you and zero pressure either way. You know I think you're great exactly as you are, but I'm completely down to be your partner in crime if you ever want to kickstart some new healthy habits!

What readers flagged

  • Does not sound human 5 readers
  • Sounds like a template 6 readers
  • Too perfect 4 readers
  • Too flattering 2 readers

A new prompt every day

Read two AI answers to the same prompt and pick the one that sounds more human. Then see how everyone else heard it.

Play today's prompt

How these numbers work

  • Both answers were written by AI models. We don't name the models while a season is running.
  • Each reader saw the two answers in random order and placed a pick on a seven-point scale. The bars count which side they leaned to.
  • Only a reader's first answer to this prompt counts.
  • These readers are people who play AI Vibe Check, so they don't stand in for a wider group. There is no correct answer. The bars show which answer read as human to them.