Is this a problem for which the (human) solution is well documented an known and was learned during the training phase? Or is it a novel problem? I personally

Is this a problem for which the (human) solution is well documented an known and was learned during the training phase? Or is it a novel problem? I personally think anthropomorphizing LLMs is a bad idea.