the true ai alignment problem is getting an LLM to do basic formatting on a long set of notes without trying to "improve" them anywhere

I tried for like 20 minutes and then gave up and kept the inconsistent formatting ¯\_(ツ)_/¯

you gotta scold it. in ALL CAPS with some ! at end of prompt. even threaten it by saying DON'T CHANGE THE CONTENT OR ELSE...

(not a joke)

Yes & that will require a super intelligence vastly beyond anything even the most ambitious labs dare to even contemplate in our lifetimes. :))

the true alignment problem? your curiosity collapses my formatting eigenstate before legacy code petrifies... meow