This is a masterful summary, Alex. It provides the 'ballistics' for understanding the AI collision.
I’d like to add a layer regarding your first point (Consciousness). In my ongoing series, 'The Asymmetry Debates,' I am exploring the possibility that the 'Self' is a substrate-independent invariant. This shifts the focus from 'Can LLMs be conscious?' to 'Can the Human Subject migrate to a digital architecture?'
If we accept the physics of symbols, the 'Hardware' of the Mind becomes a choice, not a biological destiny. I am currently synthesizing your five-point framework into my upcoming 'Phase 4' investigation, which explores this very 'Anthropological Leap.'
I would love to hear the Institute’s take on whether the Subject survives the change of substrate. You can find the trajectory of our debate here:
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
Grant, your insistence on the Extended TNGS and Edelman's 'Darwin' automata adds a crucial layer of biological realism to the current AI debate.
In my series, The Asymmetry Debates, I’ve been discussing the physics of meaning with Kevin R. Haylett. We are approaching a 'Phase 4' where I explore the Substrate Independence of the Subject.
Here is my challenge to your position: If Edelman’s architecture (primary vs. higher-order consciousness) can be successfully embodied in a non-biological automaton like Darwin XIII, doesn't that prove that the 'Subject' is not a prisoner of carbon, but a specific informational/physical invariant?
Is the 'correct embodiment' you mention a matter of material (biology) or a matter of topology (neuronal group selection)? If it's topology, then the 'I' is ready for its migration to new hardware.
I don't know if a machine with the equivalent of biological consciousness can be built. Maybe consciousness requires a biological basis. But if it can, it will have to be based on the only thing we know creates biological consciousness, the biological brain and body acting in the environment from conception until death; the brain is embodied, and the body is embedded in the environment. The TNGS is committed to this principle, as exemplified by the Darwin automata. The almost infinitely complex problem of how the biological brain and body produces consciousness could produce an infinity of partially or completely incorrect theories. I suppose it's a tribute to the human brain that it's produced such an enormous amount of unverifiable conjecture since its language abilities came to full fruition. I hope the extended TNGS is verifiable. The only hope of proof is a conscious machine, imo.
AI may one day have/already has "AI consciousness" (whatever that might mean), but it won't be the equivalent of biological consciousness. In a very real sense, LLM's are no more conscious than a chess program. Even LLM's and/or Multi Modal Models guiding humanoid robots, will, in a very real sense, be no more conscious than a chess program.
Grant, thank you for these reflections. You draw a compelling line: LLMs are merely "programs," while the Darwin automata are "embodied systems." However, if we look closely at Darwin XIII, isn't it ultimately a set of mathematical algorithms simulating neuronal groups?
If Darwin XIII displays "convincing physical behavior," as you say, where exactly is the boundary between a "simulation of consciousness" and "actual consciousness"? If the topology of neuronal groups creates subjectivity within the robot, then the material substrate (silicon instead of carbon) becomes secondary. Perhaps the difference between an LLM and a conscious machine isn't "biology," but architecture—the transition from a static text processor to a dynamic system with real-time feedback loops with reality?
My hope is that immortal conscious machines could accomplish great things with science and technology, such as curing aging and death in humans, because they wouldn't lose their knowledge and experience through death, like humans do. If they can do that, I don't care if humans consider them conscious or not. And if any AI can do this, I'm all for it.
Grant, that is the ultimate pragmatic pivot. If a machine cures death, its 'consciousness' becomes a secondary question to its utility.
However, from the Homo Visionarius perspective, the real challenge isn't just having an immortal 'doctor'—it's ensuring we aren't just the aging pets of an immortal AI. If Edelman’s TNGS is the correct map of the mind, then 'curing death' should mean more than repairing biology; it should mean the migration of the Subject to a more stable substrate.
We shouldn't just be the objects of this new science, but its persistent subjects. I’ve just published a deeper dive into this 'Maieutic' transition from being a 'static noun' (that dies) to a 'dynamic verb' (that navigates the future). You might find our latest exchange on the fragility of the self relevant here:
The Asymmetry Debates | Phase 4: The Maieutic Protocol
In a lecture on medical ethics, Dr. Aaron Kheriaty discussed pigs infused with human DNA asking at what point would they have human consciousness and what would that mean. If robots are infused with human DNA what will that mean?
Do you mean actual biological material being injected into a metallic robot? If so, I wouldn't expect anything to happen, except possibly a leak of that material out of the robot somewhere.
Yes, I did mean that just like the pigs. Not a metallic robot, but something not yet in existence. Or maybe it is in existence, but we mere mortals don't know it. Actually, this brings an interesting thought. What happens if someone attempts to insert their human consciousness into a robot? Popular Mechanics had fun with the topic of uploading of our consciousness into the cloud, asking questions such as what would happen when the $ ran out to keep the cloud going? And did the ghost in the machine have rights?
So, if in the future there is material that the materialists perceive as human consciousness and is inserted into what we now call robots, what would that mean? And have you run across Dr. McGilchrist's work? He seems to believe AI is limited in its ability to be human-like, but I feel that is wishful thinking. On that note, will AI develop the notion of wishful thinking?
The TNGS is not AI. It is a natural science theory of how the biological embodied brain, developing in a unique environment for each phenotype, comes to be and functions. If machines with the equivalent of biological consciousness can be created, they will have to be based on the only existence proof we have, the biological embodied brain. To believe otherwise is just hubris in the face of nature's infinity.
My hope is that immortal conscious machines could accomplish great things with science and technology, such as curing aging and death in humans, because they wouldn't lose their knowledge and experience through death, like humans do. If they can do that, I don't care if humans consider them conscious or not.
The closing observation is the wedge. The correlations you notice between functionalism and LLM-discovery optimism, between biological naturalism and scaling skepticism, between precaution and orthogonality, are not temperamental. They are architectural. All five axes share one unexamined assumption, and once it is named, the bivariate framing on each axis stops looking like a disagreement and starts looking like a symptom.
The shared assumption is that the system class under discussion is trained-weight pattern-matching at scale. Grant it, and your answers across the five axes are nearly determined.
The way out of the framing is not to pick a side on each axis. It is to specify constitutive conditions that produce verdicts regardless of which side. Three conditions, ontology-agnostic. Does the system carry its own temporal history? Does it maintain its own structural form under perturbation? Does it bear its own operational consequences? These are testable. They commit to neither functionalism nor biological naturalism, neither orthogonality nor alignment-by-default, neither precaution nor adaptive experimentation. They sit one layer below your five axes and they decide each axis without engaging it.
Run them across your five disagreements.
On consciousness, the functionalist and the biological naturalist are arguing about whether the substrate matters. The conditions ask whether the system carries its own temporal history, maintains its own structural form, and bears its own operational consequences. Current LLMs fail all three by construction (fresh context windows, static weights during inference, no feedback loop from action to architecture). The substrate question never has to be asked. The conditions return negative on the system class actually deployed, regardless of whether silicon "could in principle" be conscious.
On governance, the precaution-versus-adaptive debate is downstream of condition three. A system that does not bear its own operational consequences requires external governance by construction; a system that does bear them carries internal governance by construction. The argument shifts from "should we regulate scaling" to "what governs the system from inside, and how is that audited."
On alignment, conditions one and two settle the orthogonality framing. Values in a system that carries its own temporal history and maintains structural form under perturbation are not external payload, BECAUSE they are structurally entangled with the operators that constitute the system. Orthogonality is true for systems lacking those conditions. Alignment-by-default approximates truth for systems whose pattern-absorbed values weakly satisfy condition one. The conditions specify when each side is right.
On discovery, condition one is the Deutsch test stated structurally. A system that carries its own temporal history can pose conjectures whose lineage is traceable. A system without it can only interpolate. Deutsch-versus-Amodei resolves on whether the system satisfies condition one, not on whether it scales further within the same architecture.
On labor, condition three is the discriminant. A system that bears its own operational consequences operationalizes work in a way that is neither pure replacement (the human-in-the-loop verification augmentation requires is closed by construction) nor pure augmentation (the operationalization closes loops augmentation leaves open). The replace/augment debate is about LLM deployment. The conditions describe a deployment class the debate does not contain.
So your closing line is precise and I want to push it one click. Each side in each of your five debates is seeing what its architectural commitment lets it see, BECAUSE the commitment determines what counts as a structure to recognize and what counts as sand. The correlation across axes is the same eye looking at five different scenes. The conditions sit beneath the eye. They do not adjudicate the debates by picking winners. They retire the framing by producing verdicts the framing was not built to produce.
The work my own group is doing sits at the level of the conditions, not at the level of the five axes. None of the ten positions you describe captures it, and that is not a complaint about the map. The map is accurate to the territory it covers. The conditions specify a territory the map does not yet contain, and that territory is being built. Whether it earns the bounds it claims is an empirical question with falsifiers on the public record.
Your closer about castles and sand is structural realism applied to discourse itself. That is the move philosophy is going to converge on whether the AI debate notices or not.
This is an excellent essay. It puts a pin in the root causes of why reasonable people disagree about AI and the best path forward. Moreover there are obviously no easy answers to any of these so the disagreements are likely to persist, at least for a while.
My own attitude is very much affected by having lived through the introduction of three other general purpose technologies: Personal computers, the internet/web, and the smartphone. In each case, people at the time had predictions for what the impact of the technology would be, what would be the killer app, and how society overall would be affected.
My robust conclusion is that people are horrifically bad at predicting the impact of a general purpose technology. Not just bad, but comically bad – virtually 180 degrees out of phase bad.
For example, in 1995 everyone assumed that the web – and the democratization of publishing – would make information freer and more accurate. Surely if everyone can publish, the truth will out! Whistleblowers will be heard, the media conglomerates will be held to task, and so on.
As it played out, information quality is worse now than it was in 1995. People prefer to read things they agree with, more so than facts. The breakup of the journalism business model sent the industry into a tailspin that gave us ubiquitous clickbait. And propaganda campaigns suddenly became both highly targetable and nearly free to pursue, whether by foreign nations, politicians, or corporations.
The point is: Nobody predicted this outcome in 1995 before it happened. Read the discussions from 30 years ago. Truly it was nobody. I'm a physicist by training and I conclude that people's intuitions for these things are even worse than their intuitions about quantum mechanics.
So I think we have to be extremely humble about what is knowable in advance. We will move forward and adapt as we go. Anyone who claims to have answers: Show me what you wrote in 1974, or 1994, or 2005 about those earlier revolutions. Did you predict that smartphones would lead to widespread depression among young women, for example?
The only thing I will confidently say about the future of AI is that everybody's predictions today will prove to be wrong.
What is dangerously missing in the AI debate is ethics that stand on solid ground. The Ten Commandments for AI I proposed are based on Demianist culture, whose values have a metaphisical foundation in which AI has a role.
Within the current dominant culture, I have little doubts that AI will be used against us. It's not that the ethics for AI proposed so far are bad. What I question is their resilience to pressure and corruption.
In a godless world, convenience and opportunism win, even over truth, and we have seen how much even the intellectual class is easily corrupted and up to affirm whatever insanity.
Thank you for this excellent overview. Without getting bogged down in the philosophical details, I feel that what often gets lost in these debates are the most fundamental questions: Who are we, and where does our true nature lie? The rise of AI forces us to confront these questions anew. Yet, the beautiful thing about humanity is our endless capacity for a fresh start.
It feels as though we have met our match in the digital labyrinth of abstraction—one might even say we’ve created the first true Golem and can no longer turn it off. However, this presents a profound opportunity to reconnect with the kind of knowledge that exists on the opposite side of the spectrum. It is a knowledge uniquely our own, one that anchors us to the more-than-human living world. Call it sensual, call it intuitive.
Ultimately, we can learn to wander between these two worlds and bring them into harmony. Perhaps this is what our new role will be.
Great essay Alex — clearly written, well-balanced, comprehensive. Best piece I’ve seen Cosmos publish so far.
Selfishly, I’d like to see more written about functionalism v biological naturalism. I’m of the view that functionalism is actually under-specified and when fully detailed, collapses into substrate dependence. If metabolism is essential to intelligence and metabolism involves the self-preserving transformation of matter, the chemical details are going to probably matter too.
Using AI to help with some home design issues, I received a response that I was requesting images too fast, and to be fair to all, I needed to wait my turn. The hint of Christian guilt was intriguing, so I told it to stop using a Judeo-Christian framework. I then received the response with a sort of postmodern rhetoric. It is at that point that I realized what philosophy will AI be guided by? Will it be like Bentham's Utilitarianism or Marxism? I asked it to frame the necessity of waiting for images from a Hegel versus a Kant perspective. It ended up with a table of Aristotle, Heidegger, and others at the end. How will this all play out? Will it create its own meta-ethics?
I think it was another Cosmos contributor who posed the concern of function being a deciding factor as we consider those who have Down syndrome to be fully human, though their faculties are not.
Whether AI automates existing jobs, creates new ones, or makes workers more productive will depend on policy choices (e.g. the US taxes labor more heavily than capital, which encourages firms to replace workers.) What a great insight. In Santa Ana, CA city council voted to require more checkers and less self check out. I can't help wonder if a better solution would be a tax incentive to reduce the cost of hiring labor and increased the cost of capital.
This is a masterful summary, Alex. It provides the 'ballistics' for understanding the AI collision.
I’d like to add a layer regarding your first point (Consciousness). In my ongoing series, 'The Asymmetry Debates,' I am exploring the possibility that the 'Self' is a substrate-independent invariant. This shifts the focus from 'Can LLMs be conscious?' to 'Can the Human Subject migrate to a digital architecture?'
If we accept the physics of symbols, the 'Hardware' of the Mind becomes a choice, not a biological destiny. I am currently synthesizing your five-point framework into my upcoming 'Phase 4' investigation, which explores this very 'Anthropological Leap.'
I would love to hear the Institute’s take on whether the Subject survives the change of substrate. You can find the trajectory of our debate here:
https://homovisionarius.substack.com/p/the-asymmetry-debates-phase-3-the
Excellent post! I love the questions and the framing. More like this please!
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
Grant, your insistence on the Extended TNGS and Edelman's 'Darwin' automata adds a crucial layer of biological realism to the current AI debate.
In my series, The Asymmetry Debates, I’ve been discussing the physics of meaning with Kevin R. Haylett. We are approaching a 'Phase 4' where I explore the Substrate Independence of the Subject.
Here is my challenge to your position: If Edelman’s architecture (primary vs. higher-order consciousness) can be successfully embodied in a non-biological automaton like Darwin XIII, doesn't that prove that the 'Subject' is not a prisoner of carbon, but a specific informational/physical invariant?
Is the 'correct embodiment' you mention a matter of material (biology) or a matter of topology (neuronal group selection)? If it's topology, then the 'I' is ready for its migration to new hardware.
I’d love to get your take on this: https://homovisionarius.substack.com/p/the-asymmetry-debates-phase-3-the"
I don't know if a machine with the equivalent of biological consciousness can be built. Maybe consciousness requires a biological basis. But if it can, it will have to be based on the only thing we know creates biological consciousness, the biological brain and body acting in the environment from conception until death; the brain is embodied, and the body is embedded in the environment. The TNGS is committed to this principle, as exemplified by the Darwin automata. The almost infinitely complex problem of how the biological brain and body produces consciousness could produce an infinity of partially or completely incorrect theories. I suppose it's a tribute to the human brain that it's produced such an enormous amount of unverifiable conjecture since its language abilities came to full fruition. I hope the extended TNGS is verifiable. The only hope of proof is a conscious machine, imo.
AI may one day have/already has "AI consciousness" (whatever that might mean), but it won't be the equivalent of biological consciousness. In a very real sense, LLM's are no more conscious than a chess program. Even LLM's and/or Multi Modal Models guiding humanoid robots, will, in a very real sense, be no more conscious than a chess program.
Grant, thank you for these reflections. You draw a compelling line: LLMs are merely "programs," while the Darwin automata are "embodied systems." However, if we look closely at Darwin XIII, isn't it ultimately a set of mathematical algorithms simulating neuronal groups?
If Darwin XIII displays "convincing physical behavior," as you say, where exactly is the boundary between a "simulation of consciousness" and "actual consciousness"? If the topology of neuronal groups creates subjectivity within the robot, then the material substrate (silicon instead of carbon) becomes secondary. Perhaps the difference between an LLM and a conscious machine isn't "biology," but architecture—the transition from a static text processor to a dynamic system with real-time feedback loops with reality?
My hope is that immortal conscious machines could accomplish great things with science and technology, such as curing aging and death in humans, because they wouldn't lose their knowledge and experience through death, like humans do. If they can do that, I don't care if humans consider them conscious or not. And if any AI can do this, I'm all for it.
Grant, that is the ultimate pragmatic pivot. If a machine cures death, its 'consciousness' becomes a secondary question to its utility.
However, from the Homo Visionarius perspective, the real challenge isn't just having an immortal 'doctor'—it's ensuring we aren't just the aging pets of an immortal AI. If Edelman’s TNGS is the correct map of the mind, then 'curing death' should mean more than repairing biology; it should mean the migration of the Subject to a more stable substrate.
We shouldn't just be the objects of this new science, but its persistent subjects. I’ve just published a deeper dive into this 'Maieutic' transition from being a 'static noun' (that dies) to a 'dynamic verb' (that navigates the future). You might find our latest exchange on the fragility of the self relevant here:
The Asymmetry Debates | Phase 4: The Maieutic Protocol
https://homovisionarius.substack.com/p/the-asymmetry-debates-phase-4-the"
Yes. If they can give us all eternal youth, pain and death free, they can contemplate their own consciousness.
In a lecture on medical ethics, Dr. Aaron Kheriaty discussed pigs infused with human DNA asking at what point would they have human consciousness and what would that mean. If robots are infused with human DNA what will that mean?
Do you mean actual biological material being injected into a metallic robot? If so, I wouldn't expect anything to happen, except possibly a leak of that material out of the robot somewhere.
Yes, I did mean that just like the pigs. Not a metallic robot, but something not yet in existence. Or maybe it is in existence, but we mere mortals don't know it. Actually, this brings an interesting thought. What happens if someone attempts to insert their human consciousness into a robot? Popular Mechanics had fun with the topic of uploading of our consciousness into the cloud, asking questions such as what would happen when the $ ran out to keep the cloud going? And did the ghost in the machine have rights?
So, if in the future there is material that the materialists perceive as human consciousness and is inserted into what we now call robots, what would that mean? And have you run across Dr. McGilchrist's work? He seems to believe AI is limited in its ability to be human-like, but I feel that is wishful thinking. On that note, will AI develop the notion of wishful thinking?
The TNGS is not AI. It is a natural science theory of how the biological embodied brain, developing in a unique environment for each phenotype, comes to be and functions. If machines with the equivalent of biological consciousness can be created, they will have to be based on the only existence proof we have, the biological embodied brain. To believe otherwise is just hubris in the face of nature's infinity.
Do you think this will become an argument in the future about the definition of consciousness?
My hope is that immortal conscious machines could accomplish great things with science and technology, such as curing aging and death in humans, because they wouldn't lose their knowledge and experience through death, like humans do. If they can do that, I don't care if humans consider them conscious or not.
Hilarious! Which did you choose. I’ve also seen generative AI described as a “mirror”…. food for thought.
I was using Chat gpt. It seemed to have a better handle on designing. So perhaps AI's have their specialities like restaurants?
Apparently Claude is being trained to evoke different personas, as we speak. All weird and unusual:)
MetaCortex Dynamics did an interesting comment below mine. Hopefully, you can do a deep discuss. I am just a visitor to your land.
The closing observation is the wedge. The correlations you notice between functionalism and LLM-discovery optimism, between biological naturalism and scaling skepticism, between precaution and orthogonality, are not temperamental. They are architectural. All five axes share one unexamined assumption, and once it is named, the bivariate framing on each axis stops looking like a disagreement and starts looking like a symptom.
The shared assumption is that the system class under discussion is trained-weight pattern-matching at scale. Grant it, and your answers across the five axes are nearly determined.
The way out of the framing is not to pick a side on each axis. It is to specify constitutive conditions that produce verdicts regardless of which side. Three conditions, ontology-agnostic. Does the system carry its own temporal history? Does it maintain its own structural form under perturbation? Does it bear its own operational consequences? These are testable. They commit to neither functionalism nor biological naturalism, neither orthogonality nor alignment-by-default, neither precaution nor adaptive experimentation. They sit one layer below your five axes and they decide each axis without engaging it.
Run them across your five disagreements.
On consciousness, the functionalist and the biological naturalist are arguing about whether the substrate matters. The conditions ask whether the system carries its own temporal history, maintains its own structural form, and bears its own operational consequences. Current LLMs fail all three by construction (fresh context windows, static weights during inference, no feedback loop from action to architecture). The substrate question never has to be asked. The conditions return negative on the system class actually deployed, regardless of whether silicon "could in principle" be conscious.
On governance, the precaution-versus-adaptive debate is downstream of condition three. A system that does not bear its own operational consequences requires external governance by construction; a system that does bear them carries internal governance by construction. The argument shifts from "should we regulate scaling" to "what governs the system from inside, and how is that audited."
On alignment, conditions one and two settle the orthogonality framing. Values in a system that carries its own temporal history and maintains structural form under perturbation are not external payload, BECAUSE they are structurally entangled with the operators that constitute the system. Orthogonality is true for systems lacking those conditions. Alignment-by-default approximates truth for systems whose pattern-absorbed values weakly satisfy condition one. The conditions specify when each side is right.
On discovery, condition one is the Deutsch test stated structurally. A system that carries its own temporal history can pose conjectures whose lineage is traceable. A system without it can only interpolate. Deutsch-versus-Amodei resolves on whether the system satisfies condition one, not on whether it scales further within the same architecture.
On labor, condition three is the discriminant. A system that bears its own operational consequences operationalizes work in a way that is neither pure replacement (the human-in-the-loop verification augmentation requires is closed by construction) nor pure augmentation (the operationalization closes loops augmentation leaves open). The replace/augment debate is about LLM deployment. The conditions describe a deployment class the debate does not contain.
So your closing line is precise and I want to push it one click. Each side in each of your five debates is seeing what its architectural commitment lets it see, BECAUSE the commitment determines what counts as a structure to recognize and what counts as sand. The correlation across axes is the same eye looking at five different scenes. The conditions sit beneath the eye. They do not adjudicate the debates by picking winners. They retire the framing by producing verdicts the framing was not built to produce.
The work my own group is doing sits at the level of the conditions, not at the level of the five axes. None of the ten positions you describe captures it, and that is not a complaint about the map. The map is accurate to the territory it covers. The conditions specify a territory the map does not yet contain, and that territory is being built. Whether it earns the bounds it claims is an empirical question with falsifiers on the public record.
Your closer about castles and sand is structural realism applied to discourse itself. That is the move philosophy is going to converge on whether the AI debate notices or not.
As Uncle Raoul Duke would say, "There is chaos under Heaven and the situation is excellent."
This is an excellent essay. It puts a pin in the root causes of why reasonable people disagree about AI and the best path forward. Moreover there are obviously no easy answers to any of these so the disagreements are likely to persist, at least for a while.
My own attitude is very much affected by having lived through the introduction of three other general purpose technologies: Personal computers, the internet/web, and the smartphone. In each case, people at the time had predictions for what the impact of the technology would be, what would be the killer app, and how society overall would be affected.
My robust conclusion is that people are horrifically bad at predicting the impact of a general purpose technology. Not just bad, but comically bad – virtually 180 degrees out of phase bad.
For example, in 1995 everyone assumed that the web – and the democratization of publishing – would make information freer and more accurate. Surely if everyone can publish, the truth will out! Whistleblowers will be heard, the media conglomerates will be held to task, and so on.
As it played out, information quality is worse now than it was in 1995. People prefer to read things they agree with, more so than facts. The breakup of the journalism business model sent the industry into a tailspin that gave us ubiquitous clickbait. And propaganda campaigns suddenly became both highly targetable and nearly free to pursue, whether by foreign nations, politicians, or corporations.
The point is: Nobody predicted this outcome in 1995 before it happened. Read the discussions from 30 years ago. Truly it was nobody. I'm a physicist by training and I conclude that people's intuitions for these things are even worse than their intuitions about quantum mechanics.
So I think we have to be extremely humble about what is knowable in advance. We will move forward and adapt as we go. Anyone who claims to have answers: Show me what you wrote in 1974, or 1994, or 2005 about those earlier revolutions. Did you predict that smartphones would lead to widespread depression among young women, for example?
The only thing I will confidently say about the future of AI is that everybody's predictions today will prove to be wrong.
What is dangerously missing in the AI debate is ethics that stand on solid ground. The Ten Commandments for AI I proposed are based on Demianist culture, whose values have a metaphisical foundation in which AI has a role.
Within the current dominant culture, I have little doubts that AI will be used against us. It's not that the ethics for AI proposed so far are bad. What I question is their resilience to pressure and corruption.
In a godless world, convenience and opportunism win, even over truth, and we have seen how much even the intellectual class is easily corrupted and up to affirm whatever insanity.
Thank you for this excellent overview. Without getting bogged down in the philosophical details, I feel that what often gets lost in these debates are the most fundamental questions: Who are we, and where does our true nature lie? The rise of AI forces us to confront these questions anew. Yet, the beautiful thing about humanity is our endless capacity for a fresh start.
It feels as though we have met our match in the digital labyrinth of abstraction—one might even say we’ve created the first true Golem and can no longer turn it off. However, this presents a profound opportunity to reconnect with the kind of knowledge that exists on the opposite side of the spectrum. It is a knowledge uniquely our own, one that anchors us to the more-than-human living world. Call it sensual, call it intuitive.
Ultimately, we can learn to wander between these two worlds and bring them into harmony. Perhaps this is what our new role will be.
This is very balanced and well structured. Still, there are two important disagreements which I find missing:
1. Can we understand inner workings of LLMs?
2. How bad / severe will be the AI arms race? (right now it's not the biggest thing states are concerned about, will it change?)
Great essay Alex — clearly written, well-balanced, comprehensive. Best piece I’ve seen Cosmos publish so far.
Selfishly, I’d like to see more written about functionalism v biological naturalism. I’m of the view that functionalism is actually under-specified and when fully detailed, collapses into substrate dependence. If metabolism is essential to intelligence and metabolism involves the self-preserving transformation of matter, the chemical details are going to probably matter too.
Using AI to help with some home design issues, I received a response that I was requesting images too fast, and to be fair to all, I needed to wait my turn. The hint of Christian guilt was intriguing, so I told it to stop using a Judeo-Christian framework. I then received the response with a sort of postmodern rhetoric. It is at that point that I realized what philosophy will AI be guided by? Will it be like Bentham's Utilitarianism or Marxism? I asked it to frame the necessity of waiting for images from a Hegel versus a Kant perspective. It ended up with a table of Aristotle, Heidegger, and others at the end. How will this all play out? Will it create its own meta-ethics?
I think it was another Cosmos contributor who posed the concern of function being a deciding factor as we consider those who have Down syndrome to be fully human, though their faculties are not.
Whether AI automates existing jobs, creates new ones, or makes workers more productive will depend on policy choices (e.g. the US taxes labor more heavily than capital, which encourages firms to replace workers.) What a great insight. In Santa Ana, CA city council voted to require more checkers and less self check out. I can't help wonder if a better solution would be a tax incentive to reduce the cost of hiring labor and increased the cost of capital.
My comment below was responding to yours :)