I plan on ordering a copy, is there an e-book version? That would be convenient.
I am well-versed in LaRouche’s view of cybernetics and AI, etc. Yet despite all the issues, especially with formalism and algorithms, there is a curious thing in a subset of AI called GANS, or generative adversarial networks, made famous recently as the core …
I plan on ordering a copy, is there an e-book version? That would be convenient.
I am well-versed in LaRouche’s view of cybernetics and AI, etc. Yet despite all the issues, especially with formalism and algorithms, there is a curious thing in a subset of AI called GANS, or generative adversarial networks, made famous recently as the core tech behind convincing deep-fake videos and voices.
Now looking beyond the stunning quality of these, one has to back up a step popping off with the usual LaRouchian critique, the intent as digital magic, or misleading imposter, however ironic that is, being a deep-fake engine after all.
The question is like the famous ad, “is it real or is it Memorex?” But which aspect, also, the output, or the innards, or the dance between them as not quite formal logic anymore, despite operating within a digital computer-substrate.
A couple questions arise, the huge data sets, and constant background housekeeping, are trending away from the “closed system” epistemological issue.
Another point is this engine’s output is aesthetic, very pleasing compositionally. That should raise some eyebrows.
And even if it’s copying and pasting with high granularity and balance of considerations based on some statistical map, when does imitation become so good that it doesn’t matter anymore?
Genuineness aside, LaRouche side says no, its limitations implicit to its formal, logical, cybernetic axiom-based-method, are why humans are creative and AI not.
But what if this question of method is no longer the formal-digital issue he well discredited, despite the model running on a digital computer? Short of the transfinite being corrupted because computers don’t do infinite place values, heh, that’s an issue, but besides that issue, computers running AI weight networks, are so abstracted away from digital substrate, they are essentially analog with limits. Is that enough to “evolve” creations? is it squaring the circle, or something more akin to a least-action pathway finder through multi-dimensional space, big R in math notation?
There is a foundation here that echoes all across your work. Brief explanation on the tech and its suspicious echo in Socratic Method.
The GAN works by having an imposter attempt to pass off the work in progress to an inspector, hence the name containing the word “adversarial”.
An inspector checks the progressive image or other work of an imposter, and the cycle repeats recursively, and as the cycles progress, both the imposter and the inspector get better at their respective jobs.
But GAN adds a twist: the imitator teaches the inspector what it missed, acting like a one-way learning ratchet. The imposter learns as well, so they bootstrap each other!
The upwards spiral of the improving image is based on a back-and-forth dialogue; they push off each other to raise a new creation.
Just that “pushing off each other echoes all across atoms, life and society, but there is more…
The Socratic dialogue! It’s so close to the same idea as a GAN, one has to step back and re-evaluate , is it the case that the most successful AI method, other than Transformers of Chat GPT fame, is imitating a Socratic dialogue at its core?
The juicy irony of the inner adversarial arrangement, raises the question, how is that different from what we are doing when we chase down truth using the Socratic method? Or life evolving somehow in the jungle, for that matter.
It would be very useful to automatically map the networks you study based on tracing ideas through history. A real-fake detector smells like just the ticket for idea evaluation. Ideas modeled as deep-conditional relations, not static objects. Perhaps a more apt application for automation that can discern ideas from another across works, and automatically catalog their web of historical pathways, to supercharge research. Not there yet, but I smell it, despite following LaRouche for decades.
The point though is the issue of adversarial back and forth as having more universal potential than even we LaRouche folks know about, and while dangerously treading on cybernetic ground, there is a background question on what produces creative works, and how it differs from sophisticated imitation methods.
One top dog, forget his name, said the same idea as Turing, that if you can’t tell it apart from human, then it might as well be human.
The deeper irony is the GAN engine took that method of a “test” and used it to evolve images to amazing heights of realism and artistic compositions, so perhaps we need to revisit the epistemological debate on this and related questions. Perhaps there is a mean, between the two extremes?
The question of imitation is obviously front and center on most of your works, how to tell the oligarch-idea-spawn and the misleading fake idea from the one based on natural law and principles, so it’s not a peripheral issue, it seems to be THE issue. Perhaps these overlap here in the AI GAN engine a little, and this is the reason why it is so good?
An important distinction in composition, is generating the parts in such a way that they are coherent with the whole.
Whats important distinction here, is that rather than the GAN imposter getting the raw correction, “the answer from the back of the book”, if I understand it correctly, instead it tunes its generative model, so that it naturally generates images with the new principles taken into account.
Now maybe that is not perfectly correct, but it’s not a cheat sheet either, it’s changing the analog statistical weights or analog values, not merely appending an exception to a long list of conditionals. It’s dynamical more than static, but this is a grey area for me, to be further researched. The difference is important.
People live in a soup of fake everything, and discernment seems to be key pivot for being able to survive and prosper, so how to empower discernment, yes.
There is a reason why gaslighting and manipulative myth-spawning works, it attacks one’s will and ability to discern fake from true.
It’s about protecting one’s mind via wisdom, but that’s just another word for life’s hard-gained powers of discernment and judgement, really.
Yes, false axioms lead to fallacies of composition, so yes we humans are on defense- we are forced to play the inspector role or filter role, to guard our mind with reason and common sense.
The similarities of this seemingly fundamental back-and-forth proposal and its subsequent evaluation, like a dialogue, seem to be the center of how progress gets made, both for mind and the Generative Adversarial Network.
Another deeper idea is that a dialogue closes cycles of thought, thus making it periodic like a sine wave or a nested composition. Principles of Harmony once again joins the chat. The change made by feedback is also hidden in plain sight, and we know feedback makes learning and correction possible.
Would love to see where this resolves. I have a suspicion that there is more there than meets the eye, as the GAN phenomenon is demonstrating convincing wholes that are greater than the sum of its parts, by having an internal competition of sorts.
I plan on ordering a copy, is there an e-book version? That would be convenient.
I am well-versed in LaRouche’s view of cybernetics and AI, etc. Yet despite all the issues, especially with formalism and algorithms, there is a curious thing in a subset of AI called GANS, or generative adversarial networks, made famous recently as the core tech behind convincing deep-fake videos and voices.
Now looking beyond the stunning quality of these, one has to back up a step popping off with the usual LaRouchian critique, the intent as digital magic, or misleading imposter, however ironic that is, being a deep-fake engine after all.
The question is like the famous ad, “is it real or is it Memorex?” But which aspect, also, the output, or the innards, or the dance between them as not quite formal logic anymore, despite operating within a digital computer-substrate.
A couple questions arise, the huge data sets, and constant background housekeeping, are trending away from the “closed system” epistemological issue.
Another point is this engine’s output is aesthetic, very pleasing compositionally. That should raise some eyebrows.
And even if it’s copying and pasting with high granularity and balance of considerations based on some statistical map, when does imitation become so good that it doesn’t matter anymore?
Genuineness aside, LaRouche side says no, its limitations implicit to its formal, logical, cybernetic axiom-based-method, are why humans are creative and AI not.
But what if this question of method is no longer the formal-digital issue he well discredited, despite the model running on a digital computer? Short of the transfinite being corrupted because computers don’t do infinite place values, heh, that’s an issue, but besides that issue, computers running AI weight networks, are so abstracted away from digital substrate, they are essentially analog with limits. Is that enough to “evolve” creations? is it squaring the circle, or something more akin to a least-action pathway finder through multi-dimensional space, big R in math notation?
There is a foundation here that echoes all across your work. Brief explanation on the tech and its suspicious echo in Socratic Method.
The GAN works by having an imposter attempt to pass off the work in progress to an inspector, hence the name containing the word “adversarial”.
An inspector checks the progressive image or other work of an imposter, and the cycle repeats recursively, and as the cycles progress, both the imposter and the inspector get better at their respective jobs.
But GAN adds a twist: the imitator teaches the inspector what it missed, acting like a one-way learning ratchet. The imposter learns as well, so they bootstrap each other!
The upwards spiral of the improving image is based on a back-and-forth dialogue; they push off each other to raise a new creation.
Just that “pushing off each other echoes all across atoms, life and society, but there is more…
The Socratic dialogue! It’s so close to the same idea as a GAN, one has to step back and re-evaluate , is it the case that the most successful AI method, other than Transformers of Chat GPT fame, is imitating a Socratic dialogue at its core?
The juicy irony of the inner adversarial arrangement, raises the question, how is that different from what we are doing when we chase down truth using the Socratic method? Or life evolving somehow in the jungle, for that matter.
It would be very useful to automatically map the networks you study based on tracing ideas through history. A real-fake detector smells like just the ticket for idea evaluation. Ideas modeled as deep-conditional relations, not static objects. Perhaps a more apt application for automation that can discern ideas from another across works, and automatically catalog their web of historical pathways, to supercharge research. Not there yet, but I smell it, despite following LaRouche for decades.
The point though is the issue of adversarial back and forth as having more universal potential than even we LaRouche folks know about, and while dangerously treading on cybernetic ground, there is a background question on what produces creative works, and how it differs from sophisticated imitation methods.
One top dog, forget his name, said the same idea as Turing, that if you can’t tell it apart from human, then it might as well be human.
The deeper irony is the GAN engine took that method of a “test” and used it to evolve images to amazing heights of realism and artistic compositions, so perhaps we need to revisit the epistemological debate on this and related questions. Perhaps there is a mean, between the two extremes?
The question of imitation is obviously front and center on most of your works, how to tell the oligarch-idea-spawn and the misleading fake idea from the one based on natural law and principles, so it’s not a peripheral issue, it seems to be THE issue. Perhaps these overlap here in the AI GAN engine a little, and this is the reason why it is so good?
An important distinction in composition, is generating the parts in such a way that they are coherent with the whole.
Whats important distinction here, is that rather than the GAN imposter getting the raw correction, “the answer from the back of the book”, if I understand it correctly, instead it tunes its generative model, so that it naturally generates images with the new principles taken into account.
Now maybe that is not perfectly correct, but it’s not a cheat sheet either, it’s changing the analog statistical weights or analog values, not merely appending an exception to a long list of conditionals. It’s dynamical more than static, but this is a grey area for me, to be further researched. The difference is important.
People live in a soup of fake everything, and discernment seems to be key pivot for being able to survive and prosper, so how to empower discernment, yes.
There is a reason why gaslighting and manipulative myth-spawning works, it attacks one’s will and ability to discern fake from true.
It’s about protecting one’s mind via wisdom, but that’s just another word for life’s hard-gained powers of discernment and judgement, really.
Yes, false axioms lead to fallacies of composition, so yes we humans are on defense- we are forced to play the inspector role or filter role, to guard our mind with reason and common sense.
The similarities of this seemingly fundamental back-and-forth proposal and its subsequent evaluation, like a dialogue, seem to be the center of how progress gets made, both for mind and the Generative Adversarial Network.
Another deeper idea is that a dialogue closes cycles of thought, thus making it periodic like a sine wave or a nested composition. Principles of Harmony once again joins the chat. The change made by feedback is also hidden in plain sight, and we know feedback makes learning and correction possible.
Would love to see where this resolves. I have a suspicion that there is more there than meets the eye, as the GAN phenomenon is demonstrating convincing wholes that are greater than the sum of its parts, by having an internal competition of sorts.