Because they aren't. All the intelligence, alignment vs misalignment, hallucinations, conspiracy, cheating, coordination, "agency", that we see in the text gene

Because they aren't. All the intelligence, alignment vs misalignment, hallucinations, conspiracy, cheating, coordination, "agency", that we see in the text generated by large language models is a semantic projection. That the linear model of language abstracts can compress and decompress language and it can be useful is undeniable. All the contraptions built thereupon predicated on "agency" have become a societal addiction. Addiction to caffeine as oppose to alcohol might have brought about Enlightenment. Addiction to opioids is a modern tragedy that started with the private state building of British merchants. The modern addiction to the dazzling generation of human language and computer programming code by LLMs is a novel addiction and remains to be seen what impact it will have. But fundamentally, the semantic interpretation underlying this addiction is downstream from the training data compressed in the models. They are not 'lying, cheating and coordinating'. They are generating language and we're building software on top of this language and assigning meaning to the whole thing.