Im starting to believe it is exactly that...obfuscation. it started after it was known Chinese were generating models that were much smaller and more accurate f

Im starting to believe it is exactly that...obfuscation. it started after it was known Chinese were generating models that were much smaller and more accurate from the llms response. The more generally articulated the answer is along with its accuracy in what it is and not what it isn't along with many strange never written before English euphemisms that dont make sense is definitely semantical obfuscation probably added in as a post processing transform to prevent these types of actors from being able to get easy to digest answers, and leave watermarks in their models to detect.