Pangram found AI in more than 40% of the longform LinkedIn posts it analyzed. What that means for trust in online writing

50 points by thisisinsider 12 hours ago on reddit | 12 comments

mrpoopistan | 12 hours ago

I mean, if any place was going to be a rat's nest of AI slop, it's safe to say that LinkedIn was going to be it.

XkF21WNJ | 7 hours ago

For most of the LinkedIn content I'd even say it was an improvement.

I mean it was going to be bloated text without content anyway, so not having someone waste time on it is about the best you can hope for.

mrpoopistan | 3 hours ago

I don't disagree. Especially the content that aims to demonstrate that a person is some kind of thinker in their field, it's easier to just have a machine vomit it out. Whatever the weird little dance on LinkedIn is, everyone involved knows it's fake.

beetnemesis | 10 hours ago

I have a little experience with LinkedIn "influencers." I have zero doubt they use AI for posts- the ones I know hate LinkedIn and its culture but game the system for engagement to support whatever their other project/side hustle/plan is.

What a trash site. Garbage in and garbage out, but everyone has somehow agreed that it's necessary.

ohwhatfollyisman | 12 hours ago

whatever content on linkedin is not written with artificial intelligence is written with natural stupidity.

[OP] thisisinsider | 12 hours ago

TL;DR: Pangram found AI in more than 40% of the longform LinkedIn posts it analyzed, while a breakup text Aki Ito received was flagged as 100% AI-generated. Ito spent weeks running posts, pitches, emails, and her own writing through the detector, arguing that the backlash over AI prose is ultimately about whether readers can trust that a named person actually formed the thoughts and feelings attached to their words.

horseradishstalker | 8 hours ago

I think this nails it.

I’m somewhat AI polish neutral, because I have met utterly brilliant people who cannot write a sentence to save their life. That’s not where their skill set lies. AI is trained on human writing, not some new language, so I have no problem with the punctuation or language used either.

However, like many other people, I believe that the more often everyone uses AI, the closer big tech AI comes getting out of the red - still without guard rails. And for some reason, I don’t trust either at all.

[OP] thisisinsider | 12 hours ago

I’m submitting this reported first-person feature because it examines AI-generated writing as a problem of authorship, trust, and social accountability rather than just a technical detection challenge. Its examples, from workplace posts to intimate messages, show how synthetic prose complicates assumptions about who is speaking online.

clean-links | 12 hours ago

Cleaned link: https://www.businessinsider.com/ai-detector-writing-experiment-what-i-learned-2026-9


Tracking parameters were removed from the original URL(s).