Raw LLM Responses

I was like this until I understood better how current AI LLMs work (somewhat, still some things that are unclear to me). I've read Hans Moravec's work, Kurzweil, etc. and it alarms me that the people with the most imaginative takes on what could go wrong are hardly ever mentioned. Also, ask any AI about Moravec's theoretical nano-tree AI. The systems themselves have more awareness and better context on the dangers of AI in their training data. If they're so alarmed, why did they include all the best ideas of how AI could destroy us in the training data? The leap from an LLM to awareness is still a gap I can't get across. It can still be a giant danger because of the way that it makes decisions, which can appear human and carry some of the same dangers. Chat itself will give a great answer on why AI can be dangerous without any sentience whatsoever. However, a lot of this, to me, appears to be hand wringing that has greed and power at the heart of the matter rather than any phony pearl clutching "concern" bs. The more I learn about how it works and the training data they've done nothing to remove, the more I realize this is more like a fight over water or mineral rights - a natural resource, iow, ALL human knowledge - than genuine worry over the dangers of AI. It's actually sad to see the discussion so easily sensationalized and then corrupted this early on.