6 Comments
User's avatar
Popular AI's avatar

Would a Pangram “AI-written” result make you trust an article less, even if its facts and arguments held up?

Shan Mukundan's avatar

Unfortunately it is not - mostly because it is not designed to detect what matters - authorship. check out my new essay https://thesubliminal.substack.com/p/the-key-to-the-safe?r=8z5vrt

Desaree (Des) Crane's avatar

One additional observation that concerns me: after I inserted photographs into my article, Pangram's score dropped from 95% human to 77% human. The article itself did not change. I wrote every word myself, and the photographs were my own original images taken with my personal Iphone camera. They were not AI-generated, edited, or enhanced in any way.

If adding my own unedited photographs can lower the reported "human" score, it raises questions about what Pangram is actually evaluating and how reliable these percentages are as an indicator of human authorship.

Popular AI's avatar

It is theoretically possible that repetitively structured content triggers AI detection. For example, if you caption images in an AI-ish way. However, we do this all the time and it neither raised nor lowered the Pangram AI score for us.

Desaree (Des) Crane's avatar

I just read that cropping your photos sometimes will give you a lower score. I just realized I did that with one of my photos on Substack using their tools. I did it because I wanted the photo to be a certain size comparable to my other photos. Try it next time. I was floored when it happened to me today. I did not use any filters in my photos. Believe me, one of my photos of me probably should have been filtered. LOL

James Brecknell's avatar

TL;DR: Pangram is inaccurate.