Some Medium writers and editors applaud the platform’s approach to artificial intelligence. Eric Pierce, founder of Medium’s largest pop culture publication Fanfare, says he doesn’t have to rebut many AI-generated submissions and that he believes Medium’s human curators of the endorsement program lend a hand bring out the best of the platform’s human creativity. “I can’t think of a single article I’ve read on Medium in the last few months that even hints at being created by artificial intelligence,” he says. “Increasingly, Medium seems like a bastion of common sense on an internet desperate to eat itself alive.”
However, other authors and editors believe that they can still see at present plenty texts generated by artificial intelligence on the platform. Marcus Musick, a content marketing writer who edits several publications, wrote a post lamenting that the article he believed was generated by artificial intelligence. it went viral. (Reality Defender analyzed the article in question and estimated that it was 99 percent “likely manipulated.”) The article appears to be widely read and has garnered over 13,500 “claps.”
Musick believes that he not only sees possible AI content as a reader, but also encounters it frequently as an editor. He says he rejects it about 80 percent potential authors per month because he suspects they are using artificial intelligence. He doesn’t utilize AI detectors, which he calls “useless,” instead relying on his own judgment.
While the amount of likely AI-generated content on Medium is significant, the moderation challenges the platform faces – how to extract the good work and discard the garbage – are challenges that have always plagued the larger network. The AI boom has simply compounded the problem. While click farms, for example, have long been a problem, AI has given SEO-obsessed entrepreneurs a way to quickly resurrect zombie media by filling it with AI slop. There’s an entire subgenre of YouTube hustle culture entrepreneurs who create get-rich-quick tutorials that encourage others to create AI tricks on platforms like Facebook, Amazon Kindle, and, yes, Medium. (Sample headline: “Midsize AI SEO empire in one click 🤯.”)
“Medium is now in the same place as the entire Internet. Because AI content is created so quickly, it is everywhere,” says Jonathan Bailey, a plagiarism consultant. “Spam filters, human moderators and so on – those are probably the best tools they have.”
Stubblebine’s argument – that it doesn’t necessarily matter whether a platform contains a lot of garbage, as long as it is effective in amplifying good writing and limiting the reach of that garbage – is perhaps more pragmatic than any attempt to completely eliminate AI sloppiness. His moderation strategy may be the wisest approach.
It also suggests a future in which The theory of the dead Internet comes to fruition. Once the domain of extremely conspiracy thinkers online, this theory holds that the extensive majority of the Internet is devoid of real people and human-made posts, and is instead clogged with slobs and AI-generated bots. As generative AI tools become more common, platforms that give up trying to remove bots will begin to incubate an online world where human-created work will become increasingly challenging to find on platforms flooded with AI.