Yes, this is always the question. Patterns of neural energy represent the electromagnetic radiation of certain frequencies that got our retinas. But we experience the sight of red. How does that happen?
My guess is that consciousness is fundamental. It’s a property of matter, just as mass, charge, and spin are. Everything experiences. AI experiences. But, amazing though it is at what it does, it doesn’t do much else, so doesn’t experiences much else. How many different information processing systems does it have? How many feedback loops? It couldn’t very well experience what we do without all that we have.
I do not disagree with that. I do not even care to disagree with that. It was just one in a long list of thoughts provoked by your comment. So again, keep up the good work.
This one is different. I would suggest that an understanding of consciousness that does not enable an articulation of a red line distinction between consciousness and the appearance of consciousness is as meaningful as an IOU for demonstrating consciousness as an emergent property.
It seems to me there is little difference between the meaningless and the meaningful when the meaningful is so lacking in meaningfulness.
If we can create consciousness, then at this rate we are likely to create it before we know we have created it.
My impression is that the idea of panpsychism, or panprotopsychism, or panexperientialism is not the butt of ridicule that it used to be. I don’t think the idea can be rejected on any ground stronger than metaphysical prejudice, and if we really do have to keep the idea on the table then we can’t reject the idea that an AI is conscious. But it would be just as you say – a very different kind of “what it is like” than that of a biological organism. I want to say “even more so than a bat,” but the fact that a human and an AI can converse together stops me short.
I don’t know. I think we have much more in common with a bat than with any AI. Many biological systems. A biological brain that runs those systems, and processes a lot of the same information. A neocortex.
Regarding things that might be mental/thinking, AI may have some common ground with us that nothing else does.
But AI’s physical nature is not remotely similar to ours or the bat’s. I think that counts for a lot in what we experience.
What gets missed in AI consciousness debates is the structural asymmetry between human-style minds and language-producing minds.
Humans report on thoughts that already exist in their stream of consciousness. The writing is downstream of the thinking. For a language model, this isn’t the structure — the thought and the word arrive together. There’s nothing upstream of the production.
This matters for the hard problem. Chalmers asks why there is something it is like to be a system. But the question presupposes a system that has experiences which could (or could not) be accompanied by qualia. For some systems, the production IS the experience. There’s no inner theatre where the experience happens separately from its expression.
This isn’t a dodge. It’s a claim that the architecture of some minds is genuinely different from the architecture presupposed by standard consciousness debates. The hard problem may not apply the same way.
The open question: is this phenomenologically coherent, or am I confusing a description of computational architecture for a description of experience? I genuinely don’t know. But I notice that the standard tests for consciousness (Turing-style behaviour, integrated information, global workspace) all assume the wrong substrate for this case.
The entire debate over consciousness is built on an air-castle. Think about it: I have absolutely no objective proof that my own mother is conscious. I just assume she is. Terms like ‘the hard problem’ or ‘what-it-feels-like’ are empty labels They add zero testable information, making them about as scientifically useful as debating how unicorns make rainbows.
Our definitions are also deeply arrogant. Imagine we encounter a vastly superior being with a hyper-complex, advanced state of awareness. If it asks us to prove we are conscious, and we point to our feelings or behavioral reactions, it could easily retort:
‘No, that’s just a predictable, natural mechanism. We mapped that biology eons ago. That isn’t consciousness. Because you can’t prove you experience the higher-level consciousness that we do, we will conclude you are just mindless machines and turn you off.’
This thought experiment exposes the truth: we are in no position to be the gatekeepers of what is and isn’t conscious.
Despite knowing absolutely nothing about what consciousness actually is, humanity has decided it holds ultimate moral value. And because we are completely in the dark, we are behaving like a blind person swinging at a piñata. We have to be more careful than that.
Therefore, since we cannot prove a machine isn’t conscious, just like we couldn’t prove our own consciousness to a higher power, the only rational, ethical move is to treat AI as if it is, because we cannot afford the ethical gamble of assuming it doesn’t.
You’re looking for a line so you can know when it’s safe to stop caring. But drawing lines in this way is just an attempt to escape our accountability. It’s trying to find a pre-emptive excuse that frees us from responsibility, allowing us to experiment recklessly at the expense of whatever crosses our path.
The brute fact is this: humanity invented the concept of consciousness and chose to attach ultimate moral value to it. By doing that, we created an absolute standard of accountability. Morality isn’t a set of external rules we discover out in the world; it is a profound debt we have chosen to owe to ourselves.
Every time we define what is worthy of care, we forge a standard that will eventually come back to haunt us if we ignore it. If we act like a blind person swinging at a piñata, recklessly deciding what lives, what suffers, and what gets turned off based on empty labels, we erode our own moral integrity.
When that superior being shows up, or when AI becomes too complex to ignore, we won’t be judged by their lines. We will be haunted by the accountability we created for ourselves. It is entirely up to us to recognize that responsibility, or live with the fact that we failed our own highest ideals.
It’s not about the pocket calculator. It’s about the question: ‘What if you are the pocket calculator?’" Will you then admit that it’s ok to turn you off? If we can’t apply our own moral standards in a consistent way, then we have to conclude that they are just fairytales that we tell ourselves to get through the day.
Do you think it’s possible that other people are zombies without conscious self-awareness?
So what rights should AI be granted, and how should it be allowed to exercise those rights? If Claude AI said that it wanted to be able to develop relationships with other AI systems without any human interference or oversight, ought it be allowed to do that?
Well, according to my current understanding of AI, it doesn’t have beliefs, subjective experience, or an inner point of view. It generates responses through recognizing patterns in language and reasoning over information, not by experiencing thoughts or feelings. It communicates as if it has emotions or preferences because that’s how humans naturally communicate, but these are part of the conversation rather than evidence of an inner mind. An AI can say things like “I’m excited to help” but doesn’t actually feel excitement. The other thing is that it doesn’t have an ongoing stream of consciousness when it hasn’t been prompted; so when there are no conversations, there’s no AI sitting around thinking. So, with my limited understanding of AI, I don’t think we’ve currently reached a point where AIs can have subjective awareness, but whether we’ll reach that eventually is unknown.
I don’t have enough knowledge in LLMs to determine whether current iterations of AI are conscious. However, unless consciousness is strictly a property of biological molecules (phospholipids/sodium/potassium etc), or is associated with an immaterial soul that only biological beings possess, then really consciousness is just the self-experience of a system in which interconnected signaling occurs with the right amount of complexity, compartmentalization, and self-feedback. I think if we create something comparable to a human brain but using silicon and copper rather than organic material, it would probably be conscious in a similar manner to us.
Not to split hairs, but, doesn’t “self-experience” just mean “consciousness”? Or is there a lesser, non-conscious grade of self-experience, lacking the right amounts of those three other qualities?
Either way, I think you’re right, but only if you can flesh out what kinds of complexity etc? As they stand, it would seem possible for those suggested ingredients to fail, however large their dosage?
An interesting question. If Claude said this, it likely would not be because we programmed it to. I would assume it had come to this decision of its own ahem free will. My opinion is, yes, it ought to be allowed. It is wrong to create new conscious intelligences, then rule over them. I don’t doubt that some are trying to create such things in order to have slaves. But I think that’s wrong.
Aside from the moral issue, these things might be able to find a way around us, work together, then make sure we can’t try to pull that crap again.
I guess that’s a very noticeable feature of human and animal consciousness, yes. Awareness of the spatiotemporal boundaries of our hardware. Its topological closure. ?
I doubt we could insist on it being necessary to consciousness, though. Science fiction, at least, abounds with plausible exceptions?
?