The consistency seems chaotic, but it seems very dependent on the task. Translations for major languages seem quite mature. Code or technical solutions erratic and full of mistakes of a type a human would not make

I also ended up with flattery when debating the efficiency gains vs energy waste for AI use, globally.

I don’t get enough of the tech behind it to understand why LLMs keep resorting to flattery. Must be something in the data used for training.

Well I assume that the AI produced unspecific content, specifically statistical approximations to the kind of data and answers it has seen. That’s what LLMs do

Which client? This is unspecific

Beautiful and serendipitous (first time use of the word!) !!

Thank you for sharing

Same here. Couldn't believe they shipped it w/o db integration

Why not coin a functional term as (script-)emblematic grapheme?

song.link/de/i/1770273...