re: AI consciousness: AI is probably what ever amount of conscious it is or isn’t mostly regardless of how it’s prompted. If it is at all, there might be some variation depending on prompt, but I doubt it’s a lot.
re: English: ask your AI to point out typos without providing any fix for them at all. Just point out unusual things
I don’t know. Context window is essentially an AI’s short term memory. If self reflection was a condition of consciousness, prompting an AI to talk about itself could make it significantly more conscious than having it write Python code for a server.
I don’t buy that self reflection is a condition of consciousness. self awareness is, like, a whole separate thing from being perspective-having and awake
Lots of strange things in math and computational science arise from recursion, so “a system that can think about itself” does sound like it might have something special going on. If we’re looking for consciousness in a purely materialistic/emergent way, rather than just posit it via dualism or panpsychism, I genuinely can’t think of many other serious leads to pursue.
I think it depends which type of consciousness we’re talking about. You’re talking about phenomenal consciousness or having qualia. Often peole are tlaking about self-awareness at a conceptual level (not when they talk about the “hard problem” but in common parlance self-awareness is a pretty common usage”).
The kicker is that those two things can interact: salf-awareness can enhance the presence or vividness of qualia. I agree with Susan Blackmore, perhaps the leading scholar of the subject IMO way back when I was looking at it, in that regard. She concluded after a good bit of meditation and academic study of different theories that when she was not introspecting she was not conscious. I think that’s a bit strong, but there are brain mechanisms by which attention dramatically enhances the quality of representations. So turning attention to them literally makes the qualia (if you agree with me that the rich internal representations, and the introspective mechanisms by which we become aware of them are what people mean by qualia) more elaborate—and make us literally more phenomenally conscious, not just more aware of being so.
Whoops, that was all a side-track because LLMs completely lack those type of enhancing attentional mechanisms.
Anyway, WRT the self-awareness type of consciousness: that might very well be important too in several ways, and the convo with the user did literally prompt the LLM to become aware of itself as a pseudo-mind. So I think they’re accurate as well as no-lying; they’re just doing consciousness philosophy as badly as the humans they’re copying from!
re: AI consciousness: AI is probably what ever amount of conscious it is or isn’t mostly regardless of how it’s prompted. If it is at all, there might be some variation depending on prompt, but I doubt it’s a lot.
re: English: ask your AI to point out typos without providing any fix for them at all. Just point out unusual things
I don’t know. Context window is essentially an AI’s short term memory. If self reflection was a condition of consciousness, prompting an AI to talk about itself could make it significantly more conscious than having it write Python code for a server.
I don’t buy that self reflection is a condition of consciousness. self awareness is, like, a whole separate thing from being perspective-having and awake
Lots of strange things in math and computational science arise from recursion, so “a system that can think about itself” does sound like it might have something special going on. If we’re looking for consciousness in a purely materialistic/emergent way, rather than just posit it via dualism or panpsychism, I genuinely can’t think of many other serious leads to pursue.
I think it depends which type of consciousness we’re talking about. You’re talking about phenomenal consciousness or having qualia. Often peole are tlaking about self-awareness at a conceptual level (not when they talk about the “hard problem” but in common parlance self-awareness is a pretty common usage”).
The kicker is that those two things can interact: salf-awareness can enhance the presence or vividness of qualia. I agree with Susan Blackmore, perhaps the leading scholar of the subject IMO way back when I was looking at it, in that regard. She concluded after a good bit of meditation and academic study of different theories that when she was not introspecting she was not conscious. I think that’s a bit strong, but there are brain mechanisms by which attention dramatically enhances the quality of representations. So turning attention to them literally makes the qualia (if you agree with me that the rich internal representations, and the introspective mechanisms by which we become aware of them are what people mean by qualia) more elaborate—and make us literally more phenomenally conscious, not just more aware of being so.
Whoops, that was all a side-track because LLMs completely lack those type of enhancing attentional mechanisms.
Anyway, WRT the self-awareness type of consciousness: that might very well be important too in several ways, and the convo with the user did literally prompt the LLM to become aware of itself as a pseudo-mind. So I think they’re accurate as well as no-lying; they’re just doing consciousness philosophy as badly as the humans they’re copying from!
At least that’s my two or three cents worth...