@badrihippo It could be fun to see exactly how different the LLM-generated summary for an article is compared to the article itself. (I am *not* tempted to build some kind of web-based writing game where you try to get the most-wrong-possible summary. I'm not, definitely, really not...)