As social networks fill up with AI slop, trusted relationships and communities will win.

Link: AI Content Is Everywhere on Social Media, Especially LinkedIn, by Max Spero at Pangram

This is one of the core effects of AI: even when people are not engaging with AI-generated content directly, it’s hard to avoid. Our feeds are increasingly full of AI slop.

“AI-generated content appeared across all social media platforms in our data set. The average AI rate across all scanned items was 13.8%, but specific rates varied by platform and item length. On four out of five platforms, longer content was more likely to be AI-generated than shortform content. Across all platforms, one in four longform items (25.72% of items over 250 words) were fully AI-generated.”

Specifically, long-form content on LinkedIn was 41% likely to be AI-generated, which shouldn’t surprise anyone who’s browsed LinkedIn lately. Medium was 31% likely and X was 29% likely. Open social web platforms like Bluesky and Mastodon don’t seem to have been a part of the dataset, but I think it would be foolish to assume they’re immune.

Some caveats here: the analysis was done by Pangram, which builds a browser extension and back-end tech that attempts to detect AI-generated content. That’s an imperfect process, and there are no tools that are completely reliable at making this distinction. False positives and false negatives have been common with these tools, although Pangram claims a 0.01% false positive rate. So take it with a pinch of salt, but it’s reasonable to assume that these numbers are directionally true.

All of this serves to drive trust in these platforms even lower. Increasingly, people on platforms like LinkedIn are being lazy writers and using AI to produce content that you don’t want to put the effort into. I generally think that if you can’t be bothered to write something, it’s not reasonable to ask people to read it; still, there may be some value in AI assisted writing, depending on the piece and how it was produced. (That kind of AI content, by the way, was not really measured by this test.) But AI has also led to a lot of outright spam making its way into people’s feeds in order to shamelessly build clout and advertising revenue.

Both things are making these platforms unusable, which in turn is driving people to smaller communities and group chats with people they know they can trust. I believe that’s going to be a big trend: AI leading to a noticeable drop in quality that drives people away from the platforms where it’s allowed to thrive. In that world, platforms that foster trusted relationships and communities will win.

Via 404 Media, which has characteristically great coverage of the story.