Both horrific (from a user's perspective, perhaps not the underlying code).

I've noticed another type of AI slop that's prevalent in all the popular models; font sizes and variations like you wouldn't believe. It's very hard to instruct LLM's not to do this.