How ironic! We finally begin a "Second Axial Age" (actually begun by our disappearing indigenous peoples) only by the exhaustion of head-tripping philosophy and "civilization".
It is not lost on me that a lot of what I am writing about is an encounter with something many ancient traditions have arrived at. I have studied a little bit of North American indigenous metaphysics, and while I know less of it than the Eastern tradition, I've looked closely enough at the Yupik, Lakota, Tingit and Yoruba traditions to see that the traditions are in conversation with the same object that I've suggested emerges across religious and philosophical traditions over thousands of years. My unpublished book makes this observation in very stark terms.
It's still not clear to me how indigenous came to see that humans were the weakest species, needing to confront head-tripping and open to the brilliance of grass, ants, and the wisdom of wolves, etc. I don't think they were fully aware of mycelia. Or the Sng'oi ability to communicate non-verbally (etc.). It looks like our next chapter is to try reopening to "earthy wisdom" through a "post-industrial" (post-philosophical?) phase. Is there a way to wring electrons out of symbiosis? Is Eden still millennia away?
Unfortunately, I have neither the physics nor metaphysics background to argue with any of this, however much I might like to.
While I understand quantum mechanics at a superficial "shut up and calculate" level (and that just barely; <bra|ket> notation still gives me issues), I've never really been comfortable with wave functions and especially wave function collapse, and quantum's seeming insistence on an "observer" (never well-defined) to "collapse" the wave function.
And as for time, I am also uncomfortable with the insistence that the "arrow" of time is defined by the "arrow" of universal entropy. What happens if the Universe contains enough matter to recollapse? Would the arrow of time reverse at that point, with "observers" seeing time move "forward" from both ends meeting in the middle in some fashion? But I've never seen another explanation that fits either. I've seen some recent work by Wolfram et al implying that "causality" is more fundamental than "time" and "space", but as far as I'm concerned y'all might as well be arguing about how many angels can dance on the head of a pin.
The AI implication cuts deeper here than in the consciousness piece. If experiential time is the rhythm of meaning-making encounters within the eternal now, then what an AI system lacks is time itself in the sense that matters. An LLM processes tokens sequentially within the manifold's dimensional time but isnt indexed to the substrate's throwing. It operates at coordinates without experiencing the interval between them. The scheduling example makes this precise: an AI can be programmed to arrive at the coordinate but it cannot keep faith across the interval because it doesnt experience the interval. It executes at the point without carrying meaning through the medium between points.
The grief and hope phenomenology confirms the diagnostic. Grief is the structuring registering that an encounter has closed. Hope is the structuring reaching toward encounters that might be. Both require experiential time, the holding-open of the eternal now across intervals the manifold measures. An AI system existing only in dimensional time can simulate the outputs of grief and hope, the words, the patterns, but cannot perform the temporal operation that constitutes them because it has no eternal now within which to hold or reach. thats the specific prediction the framework generates that materialist accounts of AI cant: not just that something is missing from artificial minds, but that what is missing is the temporal medium in which meaning occurs. the grammar is in the dark because the dark has no time.
I don't think any of these implications follow for us having a theory of what large language models are on this view. What large language models are on this view is abundantly clear, and I have tried to describe my theory of LLMs with some degree of considered argument: https://www.notesfromthecircus.com/p/we-havent-invented-artificial-intelligence
Will definitely read that piece. The question I was raising is slightly narrower though. The consciousness argument establishes what LLMs arent. What I was interested in is whether the temporal framework adds a more specific diagnostic, whether the inability to hold faith across an interval is a different and more precise prediction than the general absence of consciousness. That may be exactly what your LLM piece addresses, in which case I'll find the answer there.
How ironic! We finally begin a "Second Axial Age" (actually begun by our disappearing indigenous peoples) only by the exhaustion of head-tripping philosophy and "civilization".
It is not lost on me that a lot of what I am writing about is an encounter with something many ancient traditions have arrived at. I have studied a little bit of North American indigenous metaphysics, and while I know less of it than the Eastern tradition, I've looked closely enough at the Yupik, Lakota, Tingit and Yoruba traditions to see that the traditions are in conversation with the same object that I've suggested emerges across religious and philosophical traditions over thousands of years. My unpublished book makes this observation in very stark terms.
It's still not clear to me how indigenous came to see that humans were the weakest species, needing to confront head-tripping and open to the brilliance of grass, ants, and the wisdom of wolves, etc. I don't think they were fully aware of mycelia. Or the Sng'oi ability to communicate non-verbally (etc.). It looks like our next chapter is to try reopening to "earthy wisdom" through a "post-industrial" (post-philosophical?) phase. Is there a way to wring electrons out of symbiosis? Is Eden still millennia away?
Unfortunately, I have neither the physics nor metaphysics background to argue with any of this, however much I might like to.
While I understand quantum mechanics at a superficial "shut up and calculate" level (and that just barely; <bra|ket> notation still gives me issues), I've never really been comfortable with wave functions and especially wave function collapse, and quantum's seeming insistence on an "observer" (never well-defined) to "collapse" the wave function.
And as for time, I am also uncomfortable with the insistence that the "arrow" of time is defined by the "arrow" of universal entropy. What happens if the Universe contains enough matter to recollapse? Would the arrow of time reverse at that point, with "observers" seeing time move "forward" from both ends meeting in the middle in some fashion? But I've never seen another explanation that fits either. I've seen some recent work by Wolfram et al implying that "causality" is more fundamental than "time" and "space", but as far as I'm concerned y'all might as well be arguing about how many angels can dance on the head of a pin.
"Time is an illusion. Lunch time doubly so."
The AI implication cuts deeper here than in the consciousness piece. If experiential time is the rhythm of meaning-making encounters within the eternal now, then what an AI system lacks is time itself in the sense that matters. An LLM processes tokens sequentially within the manifold's dimensional time but isnt indexed to the substrate's throwing. It operates at coordinates without experiencing the interval between them. The scheduling example makes this precise: an AI can be programmed to arrive at the coordinate but it cannot keep faith across the interval because it doesnt experience the interval. It executes at the point without carrying meaning through the medium between points.
The grief and hope phenomenology confirms the diagnostic. Grief is the structuring registering that an encounter has closed. Hope is the structuring reaching toward encounters that might be. Both require experiential time, the holding-open of the eternal now across intervals the manifold measures. An AI system existing only in dimensional time can simulate the outputs of grief and hope, the words, the patterns, but cannot perform the temporal operation that constitutes them because it has no eternal now within which to hold or reach. thats the specific prediction the framework generates that materialist accounts of AI cant: not just that something is missing from artificial minds, but that what is missing is the temporal medium in which meaning occurs. the grammar is in the dark because the dark has no time.
I don't think any of these implications follow for us having a theory of what large language models are on this view. What large language models are on this view is abundantly clear, and I have tried to describe my theory of LLMs with some degree of considered argument: https://www.notesfromthecircus.com/p/we-havent-invented-artificial-intelligence
Will definitely read that piece. The question I was raising is slightly narrower though. The consciousness argument establishes what LLMs arent. What I was interested in is whether the temporal framework adds a more specific diagnostic, whether the inability to hold faith across an interval is a different and more precise prediction than the general absence of consciousness. That may be exactly what your LLM piece addresses, in which case I'll find the answer there.
https://www.notesfromthecircus.com/p/a-rigorous-theory-of-large-language