The problem isn’t the use of AI. It’s the lack of editorial effort to use AI well.
Modern image generation models can handle text fine. Or the author could have left those artefacts blank and added the text themselves in “post production”
The problem is the use of AI. It’s a reliable indicator that the author doesn’t actually care about the quality of the work, so I shouldn’t bother to read the text.
This is just lazy AI use as a replacement for lazy stock image use. The details of exactly how it sucks at its job while providing something that fills a checkbox for someone who has no concern for quality are somewhat different, but the basic failure is the same.