Raw LLM Responses

I'm not going to argue in favour of current LLM consciousness (or the lack thereof actually), but I have a question: You might infer that I'm conscious based on a mix of my behaviour and some assumptions on your part. I might talk about myself and my subjective experience of consciousness, for example - that's a behaviour. And as you're likely to assume I'm human, you might conclude that I'm therefore conscious, given that as far as we know, humans experience consciousness subjectively, at least to some degree. However, I submit that I could be an AI agent, indistinguishable from a real human, communicating with you over this medium of Reddit. So far so Turing test, but what if we explicitly detach the assumption about humanity, or more precisely, challenge the assumption that only humans (or biologically embodied animals with similar brains) can be considered to be conscious? Then your claim reduces to a hard claim that LLMs *cannot* be conscious, which is a far higher bar to clear. If that's what you hold to be true, then what would need to change architecturally for LLMs to remove that constraint? I don't believe we understand consciousness fully enough to identify how it is architected in the human brain, in detail. We may have some ideas, but it still looks "emergent". LLMs are still currently simpler mechanisms than human brains, so we might have more confidence claiming that AI consciousness is impossible, but until we have a clear non-anthropocentric model of consciousness, it's just a fuzzy conclusion from the other side of the confidence curve. Rather than ask the simple question of "how do you know I'm conscious" and risk the inevitable rabbit hole that leads to, I'll ask instead: "is current AI more conscious than a dog?". Have we reached ADogGI? Is human consciousness the only game in town, in other words?