Are those hallucinations or are those novel new ideas?

We are probably one more layer of human powered improvement in technology and one layer of AI powered improvement in technology away from web singularity or the emergence of consciousness in ai systems.

Which brings another question, what is the minimum number of neurons required to create consciousness?

What are the minimum number of neurons in a dog? or in a snail? Or maybe even butterfly? A hummingbird perhaps?

Could each large language model which has a self training mechanism and exposed to different environments end up having a completely different training? and a different personality? Could they even begin to socially relate to each other, that is, to another large language model?

Could human researchers deep in this field have already seen emergence and consciousness in AI?