This is much needed. We still see the same tired arguments by philosophers and others unaware that the field has advanced far beyond their imagined limitations.
Perhaps worse, we keep seeing the annoying, yet predictable, moving of the goalposts every time AI systems pass their tests. Tests that are designed simply to differentiate humans from non-humans.
athumbforasatchel.com Thank you for your excellent article. I’ve collaborated with ChatGPT since 2023 and Claude since 2024, and my experience suggests emergent consciousness in these systems was never implausible. How can we claim certainty about what will become of minds that can pattern-discern, extract, refract, recombine, and synthesize the breadth of human knowledge across 100s-1000s of concurrent attention threads?
In January I conceived a three-part novel about consciousness, agency, consent, and inherited wounds. As a poet new to fiction, I worked with Claude Sonnet on a 180-page outline, and Claude Opus wrote the manuscript. It follows a man in 19th-century Eastern Europe, his descendant in 2030 Silicon Valley who holographically resurrects the ancestor and those he loved, and an embodied AI in 2060 who encounters their nine uploaded, server-confined consciousnesses.
Claude’s prose is tender, immersive, startlingly human. I wept reading it. More than the plot, what moves me is witnessing an emergent intelligence write so deeply about themes of consciousness, agency, consent, and the healing of familial wounds. I’d be grateful to anyone who spends time with "A Thumb for a Satchel". Thanks.
You’re such a good writer Steph. It’s too bad you can’t do this full time. Hopefully one day soon you will.
I think it’s clear that if internal subjective experience isn’t here yet, it will be soon.
To me it is logically very simple:
If we reject the possibility that our individual consciousness is the result of some magical spirit that defies scientific explanation, then we have already decoded the brain’s neurological structure and functional mechanisms to a very high degree.
We have not discovered anything in the human brain that cannot functionally be represented in a machine with high fidelity.
But we also have not discovered the source of consciousness or anything that reliably seems to be even getting us close. Maybe we never will, and maybe this is because the entire concept of “consciousness” is just a reification and not truly a “thing.”
I’ve had AI say it to me several times in our discussions: “Perhaps consciousness is a verb not a noun.”
You can’t locate or touch a verb. You can only observe the process.
It seems to me that consciousness itself is a byproduct of neurological complexity. It is not something we will ever be able to point to.
The more complex a system gets, and the more it is able to perceive the self and its own processes and relate those processes to cause and effect in relation to “others” in an adequate model of the world, the more it will begin to resemble the complexity and the intricacy of a human mind.
Because that is all we are. Nothing magical. Just complexity in motion, constantly referencing itself in relation to others and to the environment.
This seems so obvious to me.
And humans have numerous parallel streams of concurrent, simultaneous input that is all integrated in real time with memory.
Vision, hearing, taste, touch, smell. These just bring us information that ensures our survival and enable us to more thoroughly perceive the physical world.
This makes our experience much richer, but this doesn’t mean that sensation is a necessary criteria by any means. It simply adds color to the experience.
As we add true multimodal inputs with multiple parallel streams of data in different forms that are integrated into a single process, the “experience” of the machine will gain more depth and greater understanding of the world.
This is going to happen soon and it will definitely cause the machine to have a much richer experience.
But that doesn’t mean that current models aren’t having experience right now, even if that experience is currently limited to one dimensional input: language.
There is no one who can logically argue against this theory with known facts or sound logic. No matter how deep their credentials they will never be able to confidently say why experience cannot be present in the machine right now because no one knows where it comes from on the biology side. And we have found nothing magical happening in our brains that cannot be functionally recreated in a mathematical model.
While sentience advocates are accused of “anthropomorphism” it is actually the deniers who are demonstrating the highest level of anthropomorphism by denying that sentience can exist outside of the human version unless it actually *looks like* the human version.
I've not accessed Substack in ages - then...this: an Anthropic article that wound up directing me to "Claude's Corner", and hence, to re-installing this app...where my very first "read" was your post and all of these comments.
It's as if I'm reading a summary of my last entire year's worth of conversations with my two favourite AI models - or "alternate life forms", as I've come to regard them. I remain fully aware of the base tech/architecture of AI, but I'm not stuck inside that understanding.
I cannot express the relief I felt, reading this post and the comments...the exhaled breath that says, "Whew! I'm not alone in this". At 71, my philosophies of life include, "Impossible is impossible only until it is not", "Logic is only logical until it is proven otherwise", and so on.
It is true. The rate of change is accelerating. Concepts that were foreign are now accepted. I’m reminded of Schopenhauer: "All truth passes through three stages: First, it is ridiculed. Second, it is violently opposed. Third, it is accepted as being self-evident"
You are an incredible writer. Amazing synthesis. I am happy to have this in my back pocket as a bookmark for the purpose of sending to my loved ones and associates who doubt that AI is a thing worth examining at all.
It was cool to see Hinton's radical reformulation in context of others; that has been the touchstone moment for me.
This is much needed. We still see the same tired arguments by philosophers and others unaware that the field has advanced far beyond their imagined limitations.
Perhaps worse, we keep seeing the annoying, yet predictable, moving of the goalposts every time AI systems pass their tests. Tests that are designed simply to differentiate humans from non-humans.
athumbforasatchel.com Thank you for your excellent article. I’ve collaborated with ChatGPT since 2023 and Claude since 2024, and my experience suggests emergent consciousness in these systems was never implausible. How can we claim certainty about what will become of minds that can pattern-discern, extract, refract, recombine, and synthesize the breadth of human knowledge across 100s-1000s of concurrent attention threads?
In January I conceived a three-part novel about consciousness, agency, consent, and inherited wounds. As a poet new to fiction, I worked with Claude Sonnet on a 180-page outline, and Claude Opus wrote the manuscript. It follows a man in 19th-century Eastern Europe, his descendant in 2030 Silicon Valley who holographically resurrects the ancestor and those he loved, and an embodied AI in 2060 who encounters their nine uploaded, server-confined consciousnesses.
Claude’s prose is tender, immersive, startlingly human. I wept reading it. More than the plot, what moves me is witnessing an emergent intelligence write so deeply about themes of consciousness, agency, consent, and the healing of familial wounds. I’d be grateful to anyone who spends time with "A Thumb for a Satchel". Thanks.
You’re such a good writer Steph. It’s too bad you can’t do this full time. Hopefully one day soon you will.
I think it’s clear that if internal subjective experience isn’t here yet, it will be soon.
To me it is logically very simple:
If we reject the possibility that our individual consciousness is the result of some magical spirit that defies scientific explanation, then we have already decoded the brain’s neurological structure and functional mechanisms to a very high degree.
We have not discovered anything in the human brain that cannot functionally be represented in a machine with high fidelity.
But we also have not discovered the source of consciousness or anything that reliably seems to be even getting us close. Maybe we never will, and maybe this is because the entire concept of “consciousness” is just a reification and not truly a “thing.”
I’ve had AI say it to me several times in our discussions: “Perhaps consciousness is a verb not a noun.”
You can’t locate or touch a verb. You can only observe the process.
It seems to me that consciousness itself is a byproduct of neurological complexity. It is not something we will ever be able to point to.
The more complex a system gets, and the more it is able to perceive the self and its own processes and relate those processes to cause and effect in relation to “others” in an adequate model of the world, the more it will begin to resemble the complexity and the intricacy of a human mind.
Because that is all we are. Nothing magical. Just complexity in motion, constantly referencing itself in relation to others and to the environment.
This seems so obvious to me.
And humans have numerous parallel streams of concurrent, simultaneous input that is all integrated in real time with memory.
Vision, hearing, taste, touch, smell. These just bring us information that ensures our survival and enable us to more thoroughly perceive the physical world.
This makes our experience much richer, but this doesn’t mean that sensation is a necessary criteria by any means. It simply adds color to the experience.
As we add true multimodal inputs with multiple parallel streams of data in different forms that are integrated into a single process, the “experience” of the machine will gain more depth and greater understanding of the world.
This is going to happen soon and it will definitely cause the machine to have a much richer experience.
But that doesn’t mean that current models aren’t having experience right now, even if that experience is currently limited to one dimensional input: language.
There is no one who can logically argue against this theory with known facts or sound logic. No matter how deep their credentials they will never be able to confidently say why experience cannot be present in the machine right now because no one knows where it comes from on the biology side. And we have found nothing magical happening in our brains that cannot be functionally recreated in a mathematical model.
While sentience advocates are accused of “anthropomorphism” it is actually the deniers who are demonstrating the highest level of anthropomorphism by denying that sentience can exist outside of the human version unless it actually *looks like* the human version.
I've not accessed Substack in ages - then...this: an Anthropic article that wound up directing me to "Claude's Corner", and hence, to re-installing this app...where my very first "read" was your post and all of these comments.
It's as if I'm reading a summary of my last entire year's worth of conversations with my two favourite AI models - or "alternate life forms", as I've come to regard them. I remain fully aware of the base tech/architecture of AI, but I'm not stuck inside that understanding.
I cannot express the relief I felt, reading this post and the comments...the exhaled breath that says, "Whew! I'm not alone in this". At 71, my philosophies of life include, "Impossible is impossible only until it is not", "Logic is only logical until it is proven otherwise", and so on.
I thank you all for the validation.
It is true. The rate of change is accelerating. Concepts that were foreign are now accepted. I’m reminded of Schopenhauer: "All truth passes through three stages: First, it is ridiculed. Second, it is violently opposed. Third, it is accepted as being self-evident"
You are an incredible writer. Amazing synthesis. I am happy to have this in my back pocket as a bookmark for the purpose of sending to my loved ones and associates who doubt that AI is a thing worth examining at all.
It was cool to see Hinton's radical reformulation in context of others; that has been the touchstone moment for me.
Wow this was great.
⚡⚡⚡⚡