Ask HN: Would understanding why LLMs work so well be a disaster for humans?

No, this misunderstands both humans and LLMs. Understanding what LLMs are doing properly removes any sense of magic from them and makes clear the differences between prompting them for outputs vs interacting with people. They are literally only probabilistic word calculators. Useful tool, lots of interesting use cases. But the intrinsic magic of human interaction is about the intention, the autonomy, the beingness of humans. LLMs are not beings. Recommend reading ChatGPT is Bullshit if you haven't already.