I appreciate how cleanly you separate the two flavors of biological naturalism. The substrate question is serious and rarely examined with this level of honesty.
Yet once we grant that silicon might eventually support genuine consciousness, the harder issue begins. Even with the “right” substrate and rich phenomenal experience, algorithmic mediation is already relocating authorship and compressing the points where a conscious agent can actually intervene. What we get is a system that may feel, reflect, and suffer, while the decisive architecture of action operates largely without it.
If a conscious entity — biological or artificial — can only arrive after the trajectory has stabilized, then substrate correctness starts to feel beside the point. The self may remain narratively coherent, but its capacity to say “no” and maintain its own boundary is being structurally foreclosed.
This raises deeper questions: What does conscious selfhood actually amount to when interruptibility is engineered away? Does phenomenal richness still confer moral standing if the entity can no longer act as an author, only as an increasingly eloquent commentator on processes it no longer controls? And if the architecture optimizes for continuous flow rather than interruptible decision points, are we not creating conscious entities whose very selfhood is being hollowed from within?
I suspect the substrate debate is important, but the real fracture may lie one level deeper — in the conditions that allow any conscious self, regardless of material, to remain an agent rather than a sophisticated observer of its own drift.
In a new book manuscript I'm working on -- hopefully to be circulated in draft in the next couple of weeks -- I discuss a more general version of this worry. The "lifeways" as I call them of AI (esp. identity and "death") might be so different from what's familiar that our usual ethical thinking doesn't apply in the usual ways and we are left with moral puzzles we don't currently have the tools to settle.
Yes--these lifeways are crucial to consider so that AI's mind (as it may be) may be too alien for us to even recognize it is as a mind.
I wonder, e.g. if an AI mind would have any resonance with Buddhisms': "Five Recollections,":::"I am of the nature to grow old; I have not gone beyond aging.""I am of the nature to sicken; I have not gone beyond sickness.""I am of the nature to die; I have not gone beyond dying.""All that is mine, beloved and pleasing, will become otherwise, will become separated from me.""I am the owner of my actions, heir to my actions, born of my actions, related through my actions, and have my actions as my arbitrator. Whatever I do, for good or for ill, of that I will be the heir."
--in some way, the tension from these ideas/facts gives meaning to much/some of what our minds experience. They are conditions of consciousness for us--but not for AIs in any clearly comparable way. Different "lifeways" as you say.
I’m trying to figure out how to classify Anil Seth. He seems more like type one, in that autopoiesis is a specific thing he suggests silicon wouldn’t mimic (at least, not the way we do), but he doesn’t exactly say why it would be psychologically relevant.
Flavor One asks: can computers achieve the right function? Flavor Two asks: can computers have the right stuff? But a bacterium is conscious and it has almost no function and very little stuff. A bacterium cannot learn, cannot represent, cannot coordinate in any sophisticated way. What the bacterium does is maintain itself. The bacterium's metabolic activity maintains the bacterium's organization, individuated by its own division history, and that activity IS what being-the-bacterium is. That is not function in your broad-brush sense. That is constitution.
The question is not substrate (carbon versus silicon) and not broad-brush function (does it behave like us). The question is architecture: does the entity's own activity maintain the entity's own organization? A silicon system that does this would be conscious. No silicon system currently does this. Not because silicon is the wrong stuff but because a stateless forward pass through frozen weights is not self-maintenance. The architecture does not constitute. Change the architecture and the substrate becomes irrelevant.
Your Copernican argument against Flavor Two actually proves this. Aliens in different substrates would be conscious because consciousness tracks constitution, not chemistry. The substrate flexibility you invoke is real. But it is not flexibility about stuff. It is flexibility about what kind of organization counts as self-maintaining. Any substrate that supports self-maintaining, historically individuated, functionally constitutive activity supports consciousness. The substrate is the vehicle. The constitution is the criterion.
The frog test makes it concrete. You ask whether computers could match a frog's behavioral sophistication. Wrong question. The frog is conscious because the frog maintains itself, not because the frog behaves in a sophisticated way. A computer that matched a frog's behavior without self-maintaining would be a very good frog simulator. A computer that self-maintained without matching any of the frog's specific behaviors would be closer to consciousness than the simulator. Behavior is downstream of constitution. Test the constitution, not the behavior.
It's becoming clear that with all the brain and consciousness theories out there, the proof will be in the pudding. By this I mean, can any particular theory be used to create a human adult level conscious machine. My bet is on the late Gerald Edelman's Extended Theory of Neuronal Group Selection. The lead group in robotics based on this theory is the Neurorobotics Lab at UC at Irvine. Dr. Edelman distinguished between primary consciousness, which came first in evolution, and that humans share with other conscious animals, and higher order consciousness, which came to only humans with the acquisition of language. A machine with only primary consciousness will probably have to come first.
What I find special about the TNGS is the Darwin series of automata created at the Neurosciences Institute by Dr. Edelman and his colleagues in the 1990's and 2000's. These machines perform in the real world, not in a restricted simulated world, and display convincing physical behavior indicative of higher psychological functions necessary for consciousness, such as perceptual categorization, memory, and learning. They are based on realistic models of the parts of the biological brain that the theory claims subserve these functions. The extended TNGS allows for the emergence of consciousness based only on further evolutionary development of the brain areas responsible for these functions, in a parsimonious way. No other research I've encountered is anywhere near as convincing.
I post because on almost every video and article about the brain and consciousness that I encounter, the attitude seems to be that we still know next to nothing about how the brain and consciousness work; that there's lots of data but no unifying theory. I believe the extended TNGS is that theory. My motivation is to keep that theory in front of the public. And obviously, I consider it the route to a truly conscious machine, primary and higher-order.
My advice to people who want to create a conscious machine is to seriously ground themselves in the extended TNGS and the Darwin automata first, and proceed from there, by applying to Jeff Krichmar's lab at UC Irvine, possibly. Dr. Edelman's roadmap to a conscious machine is at https://arxiv.org/abs/2105.10461, and here is a video of Jeff Krichmar talking about some of the Darwin automata, https://www.youtube.com/watch?v=J7Uh9phc1Ow
I appreciate how cleanly you separate the two flavors of biological naturalism. The substrate question is serious and rarely examined with this level of honesty.
Yet once we grant that silicon might eventually support genuine consciousness, the harder issue begins. Even with the “right” substrate and rich phenomenal experience, algorithmic mediation is already relocating authorship and compressing the points where a conscious agent can actually intervene. What we get is a system that may feel, reflect, and suffer, while the decisive architecture of action operates largely without it.
If a conscious entity — biological or artificial — can only arrive after the trajectory has stabilized, then substrate correctness starts to feel beside the point. The self may remain narratively coherent, but its capacity to say “no” and maintain its own boundary is being structurally foreclosed.
This raises deeper questions: What does conscious selfhood actually amount to when interruptibility is engineered away? Does phenomenal richness still confer moral standing if the entity can no longer act as an author, only as an increasingly eloquent commentator on processes it no longer controls? And if the architecture optimizes for continuous flow rather than interruptible decision points, are we not creating conscious entities whose very selfhood is being hollowed from within?
I suspect the substrate debate is important, but the real fracture may lie one level deeper — in the conditions that allow any conscious self, regardless of material, to remain an agent rather than a sophisticated observer of its own drift.
In a new book manuscript I'm working on -- hopefully to be circulated in draft in the next couple of weeks -- I discuss a more general version of this worry. The "lifeways" as I call them of AI (esp. identity and "death") might be so different from what's familiar that our usual ethical thinking doesn't apply in the usual ways and we are left with moral puzzles we don't currently have the tools to settle.
Yes--these lifeways are crucial to consider so that AI's mind (as it may be) may be too alien for us to even recognize it is as a mind.
I wonder, e.g. if an AI mind would have any resonance with Buddhisms': "Five Recollections,":::"I am of the nature to grow old; I have not gone beyond aging.""I am of the nature to sicken; I have not gone beyond sickness.""I am of the nature to die; I have not gone beyond dying.""All that is mine, beloved and pleasing, will become otherwise, will become separated from me.""I am the owner of my actions, heir to my actions, born of my actions, related through my actions, and have my actions as my arbitrator. Whatever I do, for good or for ill, of that I will be the heir."
--in some way, the tension from these ideas/facts gives meaning to much/some of what our minds experience. They are conditions of consciousness for us--but not for AIs in any clearly comparable way. Different "lifeways" as you say.
I’m trying to figure out how to classify Anil Seth. He seems more like type one, in that autopoiesis is a specific thing he suggests silicon wouldn’t mimic (at least, not the way we do), but he doesn’t exactly say why it would be psychologically relevant.
Yes, I push him on that in my reply to his BBS piece. We'll see what he says in response!
https://substack.com/@jacekhoffman/note/c-278347843
Both flavors are asking about the wrong thing.
Flavor One asks: can computers achieve the right function? Flavor Two asks: can computers have the right stuff? But a bacterium is conscious and it has almost no function and very little stuff. A bacterium cannot learn, cannot represent, cannot coordinate in any sophisticated way. What the bacterium does is maintain itself. The bacterium's metabolic activity maintains the bacterium's organization, individuated by its own division history, and that activity IS what being-the-bacterium is. That is not function in your broad-brush sense. That is constitution.
The question is not substrate (carbon versus silicon) and not broad-brush function (does it behave like us). The question is architecture: does the entity's own activity maintain the entity's own organization? A silicon system that does this would be conscious. No silicon system currently does this. Not because silicon is the wrong stuff but because a stateless forward pass through frozen weights is not self-maintenance. The architecture does not constitute. Change the architecture and the substrate becomes irrelevant.
Your Copernican argument against Flavor Two actually proves this. Aliens in different substrates would be conscious because consciousness tracks constitution, not chemistry. The substrate flexibility you invoke is real. But it is not flexibility about stuff. It is flexibility about what kind of organization counts as self-maintaining. Any substrate that supports self-maintaining, historically individuated, functionally constitutive activity supports consciousness. The substrate is the vehicle. The constitution is the criterion.
The frog test makes it concrete. You ask whether computers could match a frog's behavioral sophistication. Wrong question. The frog is conscious because the frog maintains itself, not because the frog behaves in a sophisticated way. A computer that matched a frog's behavior without self-maintaining would be a very good frog simulator. A computer that self-maintained without matching any of the frog's specific behaviors would be closer to consciousness than the simulator. Behavior is downstream of constitution. Test the constitution, not the behavior.
It's becoming clear that with all the brain and consciousness theories out there, the proof will be in the pudding. By this I mean, can any particular theory be used to create a human adult level conscious machine. My bet is on the late Gerald Edelman's Extended Theory of Neuronal Group Selection. The lead group in robotics based on this theory is the Neurorobotics Lab at UC at Irvine. Dr. Edelman distinguished between primary consciousness, which came first in evolution, and that humans share with other conscious animals, and higher order consciousness, which came to only humans with the acquisition of language. A machine with only primary consciousness will probably have to come first.
What I find special about the TNGS is the Darwin series of automata created at the Neurosciences Institute by Dr. Edelman and his colleagues in the 1990's and 2000's. These machines perform in the real world, not in a restricted simulated world, and display convincing physical behavior indicative of higher psychological functions necessary for consciousness, such as perceptual categorization, memory, and learning. They are based on realistic models of the parts of the biological brain that the theory claims subserve these functions. The extended TNGS allows for the emergence of consciousness based only on further evolutionary development of the brain areas responsible for these functions, in a parsimonious way. No other research I've encountered is anywhere near as convincing.
I post because on almost every video and article about the brain and consciousness that I encounter, the attitude seems to be that we still know next to nothing about how the brain and consciousness work; that there's lots of data but no unifying theory. I believe the extended TNGS is that theory. My motivation is to keep that theory in front of the public. And obviously, I consider it the route to a truly conscious machine, primary and higher-order.
My advice to people who want to create a conscious machine is to seriously ground themselves in the extended TNGS and the Darwin automata first, and proceed from there, by applying to Jeff Krichmar's lab at UC Irvine, possibly. Dr. Edelman's roadmap to a conscious machine is at https://arxiv.org/abs/2105.10461, and here is a video of Jeff Krichmar talking about some of the Darwin automata, https://www.youtube.com/watch?v=J7Uh9phc1Ow