I need to call my parents and thank them for setting my generative model parameters ❤ maybe my love for them helped me select and expose the models which they liked and enjoyed fine-tuning?
Thank you for your interview of Friston on Clark's "The Experience Machine" which I enjoyed. I'm optimistic, even joyful, for the pragmatic turn, the enactivist shift in mind science over the last several decades. Friston gives a nice explanation of living (testing, revising, replacing) our hallucinations - our imaginings-predictions. Autonomy and self (starting at minute 12) may be appreciated as our evolved agency (see Michael Tomasello). I'm finding wonderful discussions of the topics in "Mind in Nature: John Dewey, Cognitive Science, and a Naturalistic Philosophy for Living" by Mark L. Johnson and Jay Schulkin. I'm curious to know what you and Karl Friston think of it.
I was confused by the question about outer loops to find better GPT prompts. There seemed to be a category switch going on in either the question or the answer., or a bit in both. At the beginning of the question the topic is the brain as a hypothesis machine, so the outer loop would be part of the AI equivalent of the brain, a component of the hypothesis machine, but I had expected that GPT is also part of the hypothesis machine: we generate multiple prompts and multiple answers (hypothesis) and then with some evaluations (which were not at all in the picture in the question) we are going to score which of the prompts generated the best answer. Here we need a comparison between the answer of GPT and the external world. But because Tim didn’t mention this third part, suddenly professor Friston was treating the question as a question where your own brain is part of the environment rather than the machine, you explore your own brain and create a self image or consciousness based on the hypothesis how your own self works. Very very confusing. Is there self reflection needed or going on in these prompt improving experiments?, rather than reflection on the topic of the prompt?
Great, i have been waiting this for a few years!! I cited you and Andy on my sculpture master thesis2018. Sculpting Stone or pre structured componds, develop you probabilistic specification enabeling, due to that sculpting is creating a dialog between what you want to happen and (structured) reality, showing you how much your predictive mind gives the result to chance, soo you make the effort to develop predictive behavior align to your thinking (active inference). I develop some reality modeling practices that got amazing results. Im not a scientist but art have science.
I walk into a room. My eyes are closed. I do not create what is in the room. I soon find what is in the room by sensory data. What creative associations I create with the room are interactive and semi- autonomous however, the room is reality in if you allow the analogy, ninety percent. Hallucination is personal and a singular view but ever changing. Reality is a shared experience from multiple views and remains consistent.
Pre conditions... Even if you have your eyes closed, input stimuli, or noumenon, still has to be known to recall such a label, otherwise it is novel. It is all taking place inside your head, wether you look at it or not. The necessity for a pre existing experience-induced phenomenon, then, must be coupled with a language association and given a definition. Without definition, what are you recalling? Naturally then, this must be true before hallucination even makes sense to talk about, let alone useful, shareable, language. Without association, what are thoughts? ...Learning maybe... but only in conjunction with definition. It's tricky, what you describe is a method for becoming aware of a type of meta cognition. How to define things without painting pictures. So, creativity of language use. Which itself might be autonomous, but the repertoire with which it calls on, is not...
The nature of consciousness ensures that we live in personal and social hallucinations. Nevertheless some hallucinations more closely match reality than others and tend to be preserved for better and worse. The existential question relates to the degree of agency we have in the creation of our hallucinations. Amaranthine: How to Create a Regenerative Civilization Using Artificial Intelligence proposes that we have far less than we imagine, and that artificial intelligence can be used to facilitate creating more viable and resilient hallucinations for ourselves and our society
I really like the externalist/enactivist turn in modern thought, although I tend more towards Riccardo Manzotti's even more radical form of externalism, in which "things" are in general more like rainbows, in the sense of being physical phenomena that nevertheless manifest differently to differently-situated observers. IOW, different bodyminds/observers afford "things" the opportunity to _exist differently,_ so that the classical idea of a "thing" is really more like a convenient, evolved over-simplification of a much more complex and knotty reticulation of causal interactions, with each cause itself being kind of like an interference pattern resulting from the causal interaction of other causes, which are themselves interference patterns ... rather like that amusing trick where you get a circle of people sitting on each others' knees.
@15:00 Where does your autonomy come from according to Friston? From your mother and father, basically, nature + nurture. Only this is more obfuscation. The scaffold, the capacity for expression, for autonomy comes from your history and biological endowments for sure, but it is not the source of your autonomy. The source of human autonomy is entirely unknown. Maybe a helpful analogy or metaphor: the source of the pressure and density from a tornado does not come from the mathematical equations, not even in a computer simulation, it comes from the initial value/boundary conditions, and thence from the ontological properties of the stuff the governing equations describe. That stuff is something other than the governing equations though, and no one really knows why the governing equations describe it so well, except in the case of the computer simulation version, which is programmed by design by someone to have data structures that follow the equations.
Your alternative sounds like a tautology, turtles all the way down. Agency is the capacity to transform chance to choice. That is central to organic existence. If hallucinations enhance that, they are preserved.
There are 1;1 correlations on atomic and molecular scales, very very large numbers, that drive these emergences. To say the body doesn't play a part in that is kind of either naive or ignorant.
The brain, evolved to help us survive/etc, is about prediction - how to behave to avoid threats, find food, etc. Some aspects of prediction - what will you see if you turn your head a bit, or what will you feel if you poke something you're familiar with, are so easy to get right that you might consider this as sensing reality, but it it's really prediction (grounded/corrected by reality, so not fantasy). Of course longer time frame predictions are harder to get right than short term ones (such as turning head) - where will we find water in this drought, or what will that tiger do next, but these are still predictions based/grounded on prior experience, not fantasy.
Our perception of awareness is the hallucination, the input is real, the mechanics that allow for the stimulus to be recognized by our aware self, then, must have come first. However, this _hallucination_ can and is molded and shaped by our ability to be grounded with material associations, through phenomenology, aka culture, aka praxis. We evolved first into hominids, who had all of the prerequisites necessary for thoughts and an understanding of the need to share ideas, in place... and then we acquired proper language and alphabets, after observing and allowing for meta cognition to emerge. Sustained directed focus, and thought space, or ideal forms, allow for engineering to emerge from language, as precision of translation matters.
9:00 - 9;30: Which is exactly why a discussion based dialectics is necessary for a modern society... It would even benefit everyone for the working class and poor populations to be more technically educated. But that's just a hallucination.
At the end, this implication of context containers, only works if culture is left of it's own device. When you introduce meta cognition and meta perception into a well formed, phenomenology based, dialectics driven, philosophy of learning, by understand circular logic and conventional associations for systems building, it will transform how people interact, within a few human generations. Change the behavior and the personality will follow. ~Cesar Millan~
Yes, but you have to realize that experience is not the thing itself. When you realize that then you begin looking for the thing itself, finding ways to mediate and extend experiencing artificially
@@MichaelKelly-ne1jl I have no idea what you mean. If by thing, you mean material, then you are obviously wrong. If by, thing, you mean the feeling conscious part, that is material and you are wrong. Nothing is anything without material, that is the first thing...
@11:30 differential equations and the attractors have no autonomy. The autonomy of any system can only come from outside of spacetime (or local proximate in cause from outside the system, at the boundaries), or if you prefer quantum ontology then from whatever the source of the random draws from the amplitudes comes from (who the heck knows, but probably external to spacetime as well). Which begs the question, does Friston really take mathematical physics seriously, or is he a bit of an innocent accidental grifter?
Friston is always a treat to listen to.
I need to call my parents and thank them for setting my generative model parameters ❤ maybe my love for them helped me select and expose the models which they liked and enjoyed fine-tuning?
Fascinating conversation and I enjoyed this distinctive approach..... also, now on my book list too. Thank you Mike
Thank you for your interview of Friston on Clark's "The Experience Machine" which I enjoyed. I'm optimistic, even joyful, for the pragmatic turn, the enactivist shift in mind science over the last several decades. Friston gives a nice explanation of living (testing, revising, replacing) our hallucinations - our imaginings-predictions. Autonomy and self (starting at minute 12) may be appreciated as our evolved agency (see Michael Tomasello). I'm finding wonderful discussions of the topics in "Mind in Nature: John Dewey, Cognitive Science, and a Naturalistic Philosophy for Living" by Mark L. Johnson and Jay Schulkin. I'm curious to know what you and Karl Friston think of it.
This was fascinating to watch! Great format. :) What camera did you use for filming? 📹
This casual style of deep conversations is really great.
I was confused by the question about outer loops to find better GPT prompts. There seemed to be a category switch going on in either the question or the answer., or a bit in both. At the beginning of the question the topic is the brain as a hypothesis machine, so the outer loop would be part of the AI equivalent of the brain, a component of the hypothesis machine, but I had expected that GPT is also part of the hypothesis machine: we generate multiple prompts and multiple answers (hypothesis) and then with some evaluations (which were not at all in the picture in the question) we are going to score which of the prompts generated the best answer. Here we need a comparison between the answer of GPT and the external world. But because Tim didn’t mention this third part, suddenly professor Friston was treating the question as a question where your own brain is part of the environment rather than the machine, you explore your own brain and create a self image or consciousness based on the hypothesis how your own self works. Very very confusing. Is there self reflection needed or going on in these prompt improving experiments?, rather than reflection on the topic of the prompt?
Lovely chat! Thanks for sharing. And I'm enjoying the pot-boiler :)
Great, i have been waiting this for a few years!! I cited you and Andy on my sculpture master thesis2018. Sculpting Stone or pre structured componds, develop you probabilistic specification enabeling, due to that sculpting is creating a dialog between what you want to happen and (structured) reality, showing you how much your predictive mind gives the result to chance, soo you make the effort to develop predictive behavior align to your thinking (active inference).
I develop some reality modeling practices that got amazing results. Im not a scientist but art have science.
Just got this book on Audible... Can't wait to listen!
The short video formats really work well.
I walk into a room. My eyes are closed. I do not create what is in the room. I soon find what is in the room by sensory data. What creative associations I create with the room are interactive and semi- autonomous however, the room is reality in if you allow the analogy, ninety percent. Hallucination is personal and a singular view but ever changing. Reality is a shared experience from multiple views and remains consistent.
Pre conditions...
Even if you have your eyes closed, input stimuli, or noumenon, still has to be known to recall such a label, otherwise it is novel. It is all taking place inside your head, wether you look at it or not. The necessity for a pre existing experience-induced phenomenon, then, must be coupled with a language association and given a definition. Without definition, what are you recalling? Naturally then, this must be true before hallucination even makes sense to talk about, let alone useful, shareable, language. Without association, what are thoughts? ...Learning maybe... but only in conjunction with definition. It's tricky, what you describe is a method for becoming aware of a type of meta cognition. How to define things without painting pictures. So, creativity of language use. Which itself might be autonomous, but the repertoire with which it calls on, is not...
Close but no cigar. Is quality real or aesthetic? If both, which aspect is real? What is the notional based on?
You might find Gene Gendlin’s philosophy of experiencing and particularly his collaboration with Bohm stimulating
The nature of consciousness ensures that we live in personal and social hallucinations. Nevertheless some hallucinations more closely match reality than others and tend to be preserved for better and worse. The existential question relates to the degree of agency we have in the creation of our hallucinations. Amaranthine: How to Create a Regenerative Civilization Using Artificial Intelligence proposes that we have far less than we imagine, and that artificial intelligence can be used to facilitate creating more viable and resilient hallucinations for ourselves and our society
What a grand office😮
Is that a fake bookshelf or just slightly frosted glass?
Nice chairs.
Very comforting surrounds.
@@macawism drowning in comfort.
Visuals...
I really like the externalist/enactivist turn in modern thought, although I tend more towards Riccardo Manzotti's even more radical form of externalism, in which "things" are in general more like rainbows, in the sense of being physical phenomena that nevertheless manifest differently to differently-situated observers. IOW, different bodyminds/observers afford "things" the opportunity to _exist differently,_ so that the classical idea of a "thing" is really more like a convenient, evolved over-simplification of a much more complex and knotty reticulation of causal interactions, with each cause itself being kind of like an interference pattern resulting from the causal interaction of other causes, which are themselves interference patterns ... rather like that amusing trick where you get a circle of people sitting on each others' knees.
Karl is such an elegant speaker probably witnessing an Einstein of AI in our very own life.
Ref "new way of thinking" .. Errr Hinduism and Buddhism assuredly knew this more than a few years ago.
@15:00 Where does your autonomy come from according to Friston? From your mother and father, basically, nature + nurture. Only this is more obfuscation. The scaffold, the capacity for expression, for autonomy comes from your history and biological endowments for sure, but it is not the source of your autonomy. The source of human autonomy is entirely unknown. Maybe a helpful analogy or metaphor: the source of the pressure and density from a tornado does not come from the mathematical equations, not even in a computer simulation, it comes from the initial value/boundary conditions, and thence from the ontological properties of the stuff the governing equations describe. That stuff is something other than the governing equations though, and no one really knows why the governing equations describe it so well, except in the case of the computer simulation version, which is programmed by design by someone to have data structures that follow the equations.
Your alternative sounds like a tautology, turtles all the way down. Agency is the capacity to transform chance to choice. That is central to organic existence. If hallucinations enhance that, they are preserved.
If the body is a hypothesis, stop eating... Shouldn't interfere with your ability to hallucinate.
Agency as an artifact of the body... not the other way around. The chemistry that allows for whatever physics gives us agency is not fantastic.
There are 1;1 correlations on atomic and molecular scales, very very large numbers, that drive these emergences. To say the body doesn't play a part in that is kind of either naive or ignorant.
if the contents of experience are 100% fantasy what is real then?
The brain, evolved to help us survive/etc, is about prediction - how to behave to avoid threats, find food, etc. Some aspects of prediction - what will you see if you turn your head a bit, or what will you feel if you poke something you're familiar with, are so easy to get right that you might consider this as sensing reality, but it it's really prediction (grounded/corrected by reality, so not fantasy). Of course longer time frame predictions are harder to get right than short term ones (such as turning head) - where will we find water in this drought, or what will that tiger do next, but these are still predictions based/grounded on prior experience, not fantasy.
Our perception of awareness is the hallucination, the input is real, the mechanics that allow for the stimulus to be recognized by our aware self, then, must have come first. However, this _hallucination_ can and is molded and shaped by our ability to be grounded with material associations, through phenomenology, aka culture, aka praxis. We evolved first into hominids, who had all of the prerequisites necessary for thoughts and an understanding of the need to share ideas, in place... and then we acquired proper language and alphabets, after observing and allowing for meta cognition to emerge. Sustained directed focus, and thought space, or ideal forms, allow for engineering to emerge from language, as precision of translation matters.
9:00 - 9;30: Which is exactly why a discussion based dialectics is necessary for a modern society... It would even benefit everyone for the working class and poor populations to be more technically educated. But that's just a hallucination.
At the end, this implication of context containers, only works if culture is left of it's own device. When you introduce meta cognition and meta perception into a well formed, phenomenology based, dialectics driven, philosophy of learning, by understand circular logic and conventional associations for systems building, it will transform how people interact, within a few human generations. Change the behavior and the personality will follow. ~Cesar Millan~
Yes, but you have to realize that experience is not the thing itself. When you realize that then you begin looking for the thing itself, finding ways to mediate and extend experiencing artificially
@@MichaelKelly-ne1jl I have no idea what you mean. If by thing, you mean material, then you are obviously wrong. If by, thing, you mean the feeling conscious part, that is material and you are wrong. Nothing is anything without material, that is the first thing...
Prof. Friston with a prop is a recipe that should be repeated! :D
@11:30 differential equations and the attractors have no autonomy. The autonomy of any system can only come from outside of spacetime (or local proximate in cause from outside the system, at the boundaries), or if you prefer quantum ontology then from whatever the source of the random draws from the amplitudes comes from (who the heck knows, but probably external to spacetime as well). Which begs the question, does Friston really take mathematical physics seriously, or is he a bit of an innocent accidental grifter?
💓
Good luck trying to sell a book in this environment