By now, I think we've all seen the deluge of AI generated articles out on the web. If you've got a keen eye, they're easy to spot. Even though we've moved away from the obvious tells - em-dashes in particular - some models like Claude still had particular "Claude-isms" which, if you know them, are very easy to spot in the wild. Think about how often you'd write "...and this is worth nailing down explicitly" in a sentence. Probably not that often. Claude overuses this kind of sentence structure regularly, making it easy to spot.
It's gone so far that whenever I read an article from any new site, I kind of drift towards Pangram to see if there's any indication of AI usage or not. If it's close to 100% AI generated, I close the page and don't bother reading the article.
I firmly believe that if ideas are worth sharing, they should be written by hand. This helps both in understanding and also cuts down on unnecessary text, given how puffed up AI generated text can be. However, I also had a phase recently where I was on the "content train", i.e. for SEO gains I really ought to be publishing once a day. This can be hard to keep up with naturally, so I drifted towards either outright AI generated articles or at the very least AI assisted articles.
I think people should be aware whether an article was done with the help of AI or not, so I've decided to tag every article according to how it was written; by me, AI or a mix of both. Some people may not be as picky as me about reading AI generated articles and so will hopefully still get some use out of those, whereas others may rightfully decide to skip over them.