It may be true (I don't know enough neuroscience to say) that LLMs & human brains use similar techniques to make connections between concepts & learn. But most humans don't speak confidently & coherently about something unless they actually know it. The ones who do... well, we have words for them.

When a chatbot gets something wrong, it’s not because it made an error. It’s because on that roll of the dice, it happened to string together a group of words that, when read by a human, represents something false. But it was working entirely as designed. It was supposed to make a sentence & it did.

I'm not aware of any single pipeline for the whole process, but if you have CCS HiFi reads, hifiasm would be my assembler of choice.