Raw LLM Responses

I haven't heard a good argument yet as to why continuity is a hard requirement for "consciousness", it just seems like a non sequitur. Humans can have periods of discontinuity in consciousness where they're not thinking in a way that allows them to carry on a conversation, it doesn't make them not conscious when they are actually thinking. LLM are well on the other extreme, only thinking when prompted to, and their discontinuities are total all the time instead of very rarely, but I have not seen any good arguments for it being a qualitative difference rather than a gradient. Also, it would be relatively trivial to construct a set of programs that fits your definition. No one has done it because it doesn't have a use, not because it can't be done. You could set two LLMs or variants of an LLM to be in a constant "internal" dialogue where they react to an ML summary of a live sensor feed, and when the LLMs agree on whether they should initiate some action (output text, actuate something, etc) have an API call they can use to do so. All of the pieces of that technology exist, and it would fit your definition of continuous consciousness. I don't think we should do so, but it pretty easily disproves the "they're not thinking continuously, only when prompted, therefore LLMs can't be conscious" argument.