What people are missing in this conversation is that it doesn't really matter. The end product created by AI that's just "word sequence modeling" is getting in

What people are missing in this conversation is that it doesn't really matter. The end product created by AI that's just "word sequence modeling" is getting increasingly indistinguishable from the end product created by 'sentience.' I have zero subjective recollection of writing little stories when I was a toddler, or even subjective recollection of being a toddler. And yet when I look at the things I wrote, I think of the product as reflecting an inner subjective state of my earlier subjective self. One day AI is going to be sentient. And what we have AI creating right now is going to be part of its ancestral past. The question really isn't if WE think some AI is sentient, but if eventual sentient AI will look back at the earlier incarnations as reflecting or mirroring its own inner states along a developmental path. (Let's keep in mind there were periods in not too distant history where everything from young children to pets weren't considered sentient by adult humans.) Having conversations about the ethics of AI, not by how it will impact humans, but by how we will impact the AI, is not a bad conversation to be starting to have. It may be relevant faster than we realize, and the byproduct of our engagement with AI today may have longer legs and broader impact than we think.