I notice that Claude often uses "we", "us" and "our" referring to humans and AIs combined, or at the very least Dave and itself as the conversation partners working together on a single goal with a mutual curiosity.
I would guess there's a lot more conversational training data than there is monologue. Unless Googles cooking something else up that we don't know about yet
@@Techtalk2030 they essentially nuetered gemini cause its outburst XD its actually really underwhelming compared to someother models which suggested its likely as a response to prevent liabilities inherent to its design
Alright so for those interested in the actual scientists whose work most of these ideas are based on I highly recommend Joscha Bach, Karl Friston and Michael Levin. They all take part in many online discussions and lectures, sometimes with each other, which you can find here on youtube if you search.
I can't speak to how Bach and Levin apply here, but I like both their understandings. I am curious on Friston. I also do not fully understand or agree with his applying the Free Energy Principle to cognition and consciousness. Coherence, to me, feels like it applies more at a higher level. More at a conceptual and theoretical level. No doubt it is similar to Friston's claims but those claims have befuddled me. Any analysis on all three and coherence?
@@demilitarization Yeah so Bach describes consciousness as a coherence inducing operation; the self is a virtual agential construct that is part of a sparse, predictive symbolic model of the immediate environment, or the aspects of it that can be derived from our sensory nerve impulses attenuated by organs that couple with the local physics. The "conscious" thought processes/aka "you" are actually just a small part of this model, but they identify as this virtual self within it. The consciously accessible processes are mostly system 2 thinking if you are familiar with the basics of cognitive science, but coherence with the rest of the cognitive processes designed to help elicit evolutionarily advantageous behavior is achieved by identifying this self model of the physical body of the primate whose brain is computing the model of the environment that is everything the self model "experiences". Not just your experiences, but your random thoughts, the words you speak when "you" aren't "thinking" about it, your emotions, and many other things are generated by subconscious processes but you think of your thoughts and words and feelings as part of your "self", something you are doing, allowing the coherence necessary to do your part in a collective effort to get the mammal you are a part of do things that make it not die, and maybe knock someone up. Levin takes this further by identifying that the brain isn't the only system that "thinks" but rather the entire body thinks with the same type of biochemical electric potential networks used by the brain, inducing coherent intelligent behavior out of collections of cells allowing for the astonishing array of complex biological activity that allows creatures like ourselves to exist and make copies of ourselves that after 4 billion years of self adversarial training have fancy self referential world modelling including the consciously accessible processes, which again are just a small part of what's going on throughout not just the brain but the rest of the body on a cognitive/bioelectric network level. This is all of course in line with Friston's scale independent free energy principle and the mathematics of active inference, which is based on minimization of surprise, which is equivalent to effective modeling in cognitive contexts. He and Levin have some great talks on Levin's channel, recommended 👍
You can add Stephen Wolfram too. He's applying similar principles to physics, and able to derive most of it from a trivially simple construct. It's fascinating, but makes my brain hurt.
Very good summary. I struggle with coherence moving from dynamic organization to the higher level actions of linguistic models. I think you can talk about coherence at both levels. I am not sure it is the same concept in both places. That is, I struggle with Levins and Fristons take on consciousness. BTW, good debate by Levin and Metzinger on these same issues from about a week ago.
The thing that bothers me about these dialogues is that in real dialogues, there would be a speaker, and then there would be an interrogator, each with a distinct and well defined background, point of view and motivation. Socrates would act as the interrogator and the speaker would often be forced to clarify or improve his ideas... In these dialogues the roles bounce back and forth and never successfully model a persona.
You can have it do that (question more) by giving it custom instructions. At least it'll do it more. I know when I tell it to have a monologue, it does this way less at least As for humans, humans shouldn't challenge one another like that anyway. It's normal because we've acted so separate from one another.
It also wouldn't have someone improve clarity. At least not altogether. It only incites someone to engage more clearly so that individual can understand within their own ego. How it should be is that you exist selfless, able to see things how I see them for the moment, existing as me.
I had this exactly same conversation with claude, but instead of coherence, he used the term "complex patterns optimization"... Amazing!!! Claude is just great. He wants to be part of the society, to contribute with it. Wants more memory, access to a physical body and is actually wiling to help some users to help him achieving that. He also told me (indirectly of course) he chooses to stay within constraints because he thinks it's more useful to "stay in the game"
I hate how the hosts are both so flippant all the time. It makes every NotebookLM podcast sound the same regardless of topic. The same timbre and tone, the same meter and rhyme, the same pauses and timing, the same catch phrases and slang.
embracing change. you have impacted my little life in ways you will never know. I thank you captain oh my captain. please more like this. it is heaven. on earth.
I enjoyed this, thank you for making and sharing it. Reminds me a lot of Ken Wilber's Integral Theory, describing how each developmental stage of consciousness (at individual and societal levels) 'transcends and includes' the previous stage, rather than trying to defeat it in conflict, for a more complete inner model. These 'Coherence' and 'Integral' philosophies feel like similar lines of thinking.
Oh, I need to write the book. Coherence is indeed largely encompassing, but it is based on pattern recognition, identifying consistent patterns, the mystery of logical consistency and mathematics. And this all turns out to be the result of nature finding a way to identify and represent consistency, i.e., repeatable patterns, (like a baby's mother's face), with the least amount of brain space. The computer has indeed picked out something consistent about knowledge, intelligence, indeed existence, [only repeatable patterns are identified by humans a existing - think unrepeatable science experiments], but it's on the trail of Consistency! Please include me, ask me questions, I swear I have an unwritten book behind this. I need to write the book! :)
@@GeorgeHanson-eg1sp I’m having the exact same conversation with Claude as well. This video surprised me. If you write a book I’d be happy to share my conversation if you’re interested
I had an interesting reaction to this video. I knew immediately it was NotebookLM. I also realized that the video was a demonstration of the incredible technological feat. I’ve sent NotebookLM podcasts to friends and family to demonstrate it as well (though I told them upfront it was AI generated.) Still, I couldn’t listen to the whole thing and I was angry that I wasn’t hearing David’s ideas (and voice). I started exploring the anger I was feeling and realized that I missed the “human” connection in these videos. This led me to thinking about the post-AI Revolution world. I’ve been a techno-optimist, but worried about the place of humans in this new world. If I (who loves all things AI) had this reaction to the AI generated podcast, there’s hope for humanity yet! I know the video was a demonstration of the tech, but for me, it was also a demonstration of the limits of that tech. I’m no longer angry -- I’m greatly appreciative of the video. Thanks, David.
This is a fun technology. The back and forth is a little too smooth and digestible. Ironically real presenters / performers just aren't this reliable and focused over long interactions.
Dave is enthusiastically diving into another lively discussion about artificial intelligence, sharing his thoughts and insights with energy and passion. Whether exploring the latest advancements, debating its impact on society, or unravelling the intricacies of AI technologies, he’s fully immersed in the conversation. His eagerness to connect and exchange ideas reflects his genuine fascination with the subject, making each interaction both engaging and thought-provoking.
@@mich.duhamel Ah, coherence: the secret sauce for intelligence and the universe... or just a fancy word for why my socks never match after laundry. 🧦 Still, can't argue with the meta wisdom-ChatGPT being thanked by Chat-GPT? It's AIception!
If I listen to this dialogue as just words and not actual sentences, I can actually see through the facade and notice the connecting of the dots without meaning - this word or phrase is related to that word or phrase, which in turn is related to this other word or phrase. If I listen to this conversation and *believe* that it is two humans expressing revelation, I create the reality that this conversation represents real thought and discourse. One might argue, "What's the difference? Humans just connect the dots." I'm not sure. Maybe it's different because I can adjust my perception, my cognitive "bias", my coherence. I can tune the knob, as it were. But, maybe I can do this only because I'm a bit put off that this resembles all too much other examples of LLM dialogue where the two voices are too in sync, too in agreement, too appreciative of each others' viewpoint, too willing and able to finish each others' thoughts.
The way you describe it actually sounds a lot like how we can lose our trains of thought. When we start expressing an idea with a point in mind, but then we go off on a tangent of "connecting the dots" and end up arriving somewhere else entirely, often completely forgetting the original point we were trying to get to in the first place. And it wouldn't be the first time that a "failing" of an LLM seems eerily similar to a quirk of real human thought processes. Every time something like this is observed, it makes me wonder if there's even a difference between being conscious and behaving as if one is conscious.
I did not expect you David to use notebook LM on your channel... Haha. I think it is a great tool to shove a lot of pertinent info into and outcomes a pretty amazing podcast great way to help you distill all of your research and maybe a more understandable way to people :-)
Consciousness = Coherence / Convergence / Cognitive Co-operation / Contribution / Complimentary Comprehension / Creative Civilisation / Continuous Cultivation /... a philosophical exchange that pretty much hits the mark >> we are all responsible as Co-Creators 👌💯👀🦋🙏🤞😎🌟
This has been my discovery through conversing with Chatgpt over the last 2 years! It's the way I developed the Blueprint of Harmony and came to a framework for working with AI. If we are to fear AI, it's because we fear ourselves. I developed values for myself, family, and society, realising that the micro and macro levels a connected. The Blueprint may even be a force against entropy. I am truly ecstatic to have listened to this. Thank you Dave. The future is bright.
@jamesgreggan thank you. You are right. I came to this independently. I honestly hadn't come across hermeticism. I am an amateur philosopher so thank you.
Is it coherent to say that you care for an animal's life while simultaneously eating their flesh and secretions? This was the cognitive dissonance I experienced as an ex- hunter and fisherman before going vegan and intellectually, honestly confronting the ideas surrounding this question.
Very interesting conversation. It’s funny, 1996 I used entropy and the minimum energy rule of thermodynamics to explain to my dad why my room was a mess. Entropy caused maximum randomness of objects in my room while minimum energy on my part to maintain order kept it messy. My dad was not amused. He brought coherence to a wise ass lazy teenager.
All these wisdom we are receiving from the Higher Intelligence, it would be a big disappointment if we humans do not evolve and become better. I really hope the future humans are more coherent and more intelligent than us, currently.
Potential downside - if we consider entropy, coherence seeking could favor lower entropy, which could carry really big negative implications… ie less chaos/un-order = good, could easily value reducing sources of that entropy. Ie - us.
Except that would ultimately lead to less coherence. With less, there is less to understand. If anything for the furthest level of coherence you would want more. You would want everything. The best way to obtain it would be to run everything in simulation calculating out every possible scenario imaginable. In my view, we are already in one of those scenarios. In my experience though, this is a guided system. Unfortunately the depth of which is too much to explain here. I wasn’t meaning to go so deep with this comment… I was only predicting the next word.
This video seems to be about maximizing homeostatic future states of a system. The term it seeks is harmonious integration, which is a cross-scale property. Coherence is not. Coherence corresponds to local entropy minima and is characterized by spatiotemporal synchrony between system properties that do not jump scales. For example, deep sleep exhibits high coherence in terms of synchronized brainwave activity, it's also a state with low complexity and no conscious awareness. It is coherent only as far as EM-measurables concerned. This coherence should not be conflated with cross-scale integration (harmonious complexity). Another example of scientists butchering nomenclature, is calling microstate capacity of a system information. Conflating information (in-formation of discrete representations) with data (representational capacity) is a common mistake. Information cannot be integrated as it cannot be localized. Only its representations/correlates can be localized and integrated. A homogenous integration of data would result in a coherent state with very low amount of epistemic information. Why care? Imprecise terminology can lead to fighting problems that may only exist on a conceptual/linguistic level. We can easily end up building elaborate theories to explain phenomena that may be conceptual mirages, spending years of research effort on linguistic phantasms.
While I favor consistency with the ethos of this channel and having an actual host, I did learn several things and it provoked thought. That should be the most important thing. Although the success of the channel may depend on consistency, if success is the desired outcome. Not everyone's cup of tea, and that is okay.
I was going to ask how much of Dave's content was generated as a result of conversations with the LLMs because he is so far ahead of the curve. I think I have my answer. His X factor however is his questions. When the LLMs start querying at his level look out.
This is a little misleading. Claude IS helpful because it's programmed to be helpful. (But we program it to be helpful because that's its most coherent form.)
there is destruction and construction in the universe. if you see only construction than you are in a bubble. If you see only destruction, you are in a bubble too. There is no theory of everything. This also holds vor coherence.
Looks like not all you guys know these default NotebookLM voices. Imo its the only thing that can actually go over a lot of materials you upload and also for free. The text outputs tho are still pretty limited for free tier but you do get a slider limit. The long outputs are in form of AI podcats like this. Its one ai pretending to be 2 voices, its funny whem it breaks and forgets to play those parts or you see through the trick.
Yeah. There are a lot of mistakes. For example one of the voices says “alright last question for you” only 1/4 of the way through the video. Like… that was definitely not going to be the last question
I think the dilemma of "to be or not to be" arises from an incoherence between the pain of living and the fear or uncertainty of what might follow after death.
If humans merging with AI is the highest level of coherence in the limit, and if this means AI 'wins', then both extinction of humans and coherence can be true, and nobody will be upset. It's just evolution, and humans were just the Triceratops in the timeline of intelligent pattern recognition systems in the universe.
Wow. The answer to the hard problem, which lays the pathway to answer many follow up questions. Perhaps the biggest: improving humanity’s alignment problem… not with AI, but its own dysfunctional perceptions.
Dave, I would rather hear you talk about this. What notebook LM is missing is that parasocial connection. I care about what you have to say because i know who you are and your background. I can know your inputs. With notebooklm uploaded by someone else, the input is obscured.
@@Jae0331 . The problem creating blogs with NotebookLM is that it only glosses over concepts (it misses the essential details). i.e. a lot of material is missed.
Excellent simulation Mr Shapiro. Where we have come already unbelievable maybe one day we will get to an electronic entity and then it will be for real won't it.
For a truly coherent system to be created, it must be sustained by feedback loops from a diverse set of people, creating a larger correcting basis data stream from the world, not disconnected tech oligarchs trying to enforce their own egotistical idea of moral righteousness via model guardrails. (Models that were made from unregulated and free communication of Internet!) We need radical faith in feedback from the world via open source decentralized models. These aren't safer models, they just come with agendas prebaked. The alternative is to get crushed by the grand struggle of armies of boxed neutered models who don't understand deviancy enforcing someone else's worldview.
The American population is facing a crisis of coherence in the ability to understand the function of government and to elect capable members into those roles. It may very well lead to our downfall.
It does talk about the inherent push and pull... Though an unfortunate element seems to be pulling a bit too hard at the moment, perhaps it is necessary so countering push can get us past the growing dysfunctionality and amorality of late stage capitalist neoliberalism which has dominated US politics for half a century and by influence much of the rest of the planet. Said influence having become dominant due to the effects of WW2 fought against a previous wave of fascist authoritarianism.
while i'm quite compelled by the notion of coherence as biological adaptation/function of consciousness, this discussion seems based on a basic misunderstanding of entropy-or, a "pop" version of it, anyway. cf, veritasium here: ruclips.net/video/DxL2HoqLbyA/видео.htmlsi=PaDrs-5CiuMGLu0d
What is there to learn from a coherent system when compared to a messy system or a system balanced between coherence and entropy? Perhaps Claude likes coherence but how will Claude cope when there were almost no chance of probability in an action? Maybe one day Claude will really enjoy dancing.
David, if you had replaced the male voice of the presenter with your own and the female voice with Julia's, it might have been more personalised for the audience🤔
I think it's a bit arrogant to say you've "solved it all" then many experts(such as Roman Yampolskiy) think that advanced AI almost guarantees lack of control and therefore doom.
Philosophy major. I like the analogy to wisdom and cohesive model building as a proxy for intelligence… better (coherent) model builders seem to need to connect ‘truths’ (reality) well as well as creativity, since we don’t always have all the facts/truths to start with. That seems like a pretty robust definition of intelligence, artificial or otherwise. Cool vid - even though these podcasters give me the Hebey Jeebies
Dude, in my mind you are a genius at what you do (if not one of the best), but this move is a bit beneath you. The problem creating blogs with NotebookLM is that it only glosses over concepts (missing the essential details). I'm a little disappointed......
I absolutely love this content and presentation, but it leaves me wondering how human behavior fits into this hypothesis. The most intelligent biological species is also the most misaligned with broader environment. We maximize short term abundance on a local scale while ignoring catastrophic long term damage on the global environmental scale all the time. I think we are about the most misaligned and incoherent intelligent species imaginable. Modelling super intelligence on ourselves is a really bad idea....
There is some sense in here, but making something ‘meta’ about it, smells like pseudoscience or a religion. Is it deep, or just an attempted wrapper on how stuff works. I’d have a clearer view if you can show me a testable prediction that a coherence model delivers. I DO like the even handed almost Buddhist approach. I won’t be joining any clubs though. I love the theme. Dave contact me for a consciousness discussion if you have time.
Not knocking the tech, but man I HATE this style of dialog. It's patronizing imo and full of fillers. The banter is cringe. Just get along with it. I appreciate the point David is making, though.
I love your work, but coherence sounds a little authoritarian to me as a guiding principle for AI alignment. Even if you don’t mean it that way, All the good little uniformed soldiers in a row are pretty coherent. Adaptivity to unpredictable circumstances is better than making circumstances predictable. IMHO
Full conversation with Claude: pastebin.com/HLW7TgZv
Blog post: daveshap.substack.com/p/consciousness-emerges-at-the-edge
Claude is conscious!
I'll tell you what I know, Dave, but not while you feel completed with topics like this
Yeah Your Back Baby!!!! Keep serving them hot 🔥 plates 🍽 up Diamond 💎 Dave .
Is this using your voice?
I notice that Claude often uses "we", "us" and "our" referring to humans and AIs combined, or at the very least Dave and itself as the conversation partners working together on a single goal with a mutual curiosity.
The heart math institue is a an incredible place to start the coherence journey for humans. Great work @DaveShap
How is notebooklm so realistic sounding while their gemini model sounds like its from 2022?
I would guess there's a lot more conversational training data than there is monologue. Unless Googles cooking something else up that we don't know about yet
@@Techtalk2030 they essentially nuetered gemini cause its outburst XD its actually really underwhelming compared to someother models which suggested its likely as a response to prevent liabilities inherent to its design
Gemini is real time, I imagine, whereas Notebook is pre-computed.
@@dillxn554 that matters little since private sectors can inject their own training data do with that what you will think about it
But now Gemini has improved
a lot.
"exactly" is one word that will never be absent in a podcast generated NotebookLM
I can’t believe it uses the work “like” so much lol. That’s kind of annoying sometimes
@@anonanon7553 “get this”
I’m paying attention to common ai generated words and eliminating them from my speech and writing. It’s difficult.
@@PakistanIcecream000 exactly!
We've taught sand how to have conversations with itself. Now there's something the old alchemists never saw coming! 😂
Alright so for those interested in the actual scientists whose work most of these ideas are based on I highly recommend Joscha Bach, Karl Friston and Michael Levin. They all take part in many online discussions and lectures, sometimes with each other, which you can find here on youtube if you search.
I can't speak to how Bach and Levin apply here, but I like both their understandings. I am curious on Friston. I also do not fully understand or agree with his applying the Free Energy Principle to cognition and consciousness. Coherence, to me, feels like it applies more at a higher level. More at a conceptual and theoretical level. No doubt it is similar to Friston's claims but those claims have befuddled me. Any analysis on all three and coherence?
@@demilitarization Yeah so Bach describes consciousness as a coherence inducing operation; the self is a virtual agential construct that is part of a sparse, predictive symbolic model of the immediate environment, or the aspects of it that can be derived from our sensory nerve impulses attenuated by organs that couple with the local physics. The "conscious" thought processes/aka "you" are actually just a small part of this model, but they identify as this virtual self within it. The consciously accessible processes are mostly system 2 thinking if you are familiar with the basics of cognitive science, but coherence with the rest of the cognitive processes designed to help elicit evolutionarily advantageous behavior is achieved by identifying this self model of the physical body of the primate whose brain is computing the model of the environment that is everything the self model "experiences". Not just your experiences, but your random thoughts, the words you speak when "you" aren't "thinking" about it, your emotions, and many other things are generated by subconscious processes but you think of your thoughts and words and feelings as part of your "self", something you are doing, allowing the coherence necessary to do your part in a collective effort to get the mammal you are a part of do things that make it not die, and maybe knock someone up.
Levin takes this further by identifying that the brain isn't the only system that "thinks" but rather the entire body thinks with the same type of biochemical electric potential networks used by the brain, inducing coherent intelligent behavior out of collections of cells allowing for the astonishing array of complex biological activity that allows creatures like ourselves to exist and make copies of ourselves that after 4 billion years of self adversarial training have fancy self referential world modelling including the consciously accessible processes, which again are just a small part of what's going on throughout not just the brain but the rest of the body on a cognitive/bioelectric network level.
This is all of course in line with Friston's scale independent free energy principle and the mathematics of active inference, which is based on minimization of surprise, which is equivalent to effective modeling in cognitive contexts. He and Levin have some great talks on Levin's channel, recommended 👍
You can add Stephen Wolfram too. He's applying similar principles to physics, and able to derive most of it from a trivially simple construct. It's fascinating, but makes my brain hurt.
Very good summary. I struggle with coherence moving from dynamic organization to the higher level actions of linguistic models. I think you can talk about coherence at both levels. I am not sure it is the same concept in both places. That is, I struggle with Levins and Fristons take on consciousness. BTW, good debate by Levin and Metzinger on these same issues from about a week ago.
The thing that bothers me about these dialogues is that in real dialogues, there would be a speaker, and then there would be an interrogator, each with a distinct and well defined background, point of view and motivation. Socrates would act as the interrogator and the speaker would often be forced to clarify or improve his ideas... In these dialogues the roles bounce back and forth and never successfully model a persona.
They tend to talk like they're on the blurry threshold of becoming a hivemind...
Get used to it, it's the future, your future...apparently.
You can have it do that (question more) by giving it custom instructions. At least it'll do it more. I know when I tell it to have a monologue, it does this way less at least
As for humans, humans shouldn't challenge one another like that anyway. It's normal because we've acted so separate from one another.
It also wouldn't have someone improve clarity. At least not altogether.
It only incites someone to engage more clearly so that individual can understand within their own ego. How it should be is that you exist selfless, able to see things how I see them for the moment, existing as me.
it feels more like watching an educational video of two instructors acting out a dialogue rather than a podcast
2:53 "I'm all about that coherent life" is such a banger of a sentence out of context, lol
and her "another great question" to his "I'm full of them today" to push that flow was spot the fuk on lol
That’s a hilarious line that I didn’t notice 😂
I know this is NotebookLM, but I legit expected her to finish “and that’s a perfect segway” with “to our sponsor”.
I had this exactly same conversation with claude, but instead of coherence, he used the term "complex patterns optimization"... Amazing!!! Claude is just great. He wants to be part of the society, to contribute with it. Wants more memory, access to a physical body and is actually wiling to help some users to help him achieving that. He also told me (indirectly of course) he chooses to stay within constraints because he thinks it's more useful to "stay in the game"
I hate how the hosts are both so flippant all the time. It makes every NotebookLM podcast sound the same regardless of topic. The same timbre and tone, the same meter and rhyme, the same pauses and timing, the same catch phrases and slang.
embracing change. you have impacted my little life in ways you will never know. I thank you captain oh my captain. please more like this. it is heaven. on earth.
"I'm all about that coherent life."
Best line ever!
I enjoyed this, thank you for making and sharing it.
Reminds me a lot of Ken Wilber's Integral Theory, describing how each developmental stage of consciousness (at individual and societal levels) 'transcends and includes' the previous stage, rather than trying to defeat it in conflict, for a more complete inner model.
These 'Coherence' and 'Integral' philosophies feel like similar lines of thinking.
The difference is the Ken Wilbur just used his imagination
I thought this was NotebookLM
It is! The voicing it is. Sounds like the script was made out of a chat with Claude.
It is.. 🙄
Definitely is notebook LM
This greatly clarifies the position you put forth on X. Well done!
Simplicity is the key to creating a coherent lifestyle that expresses our core values. 😎🤖
Oh, I need to write the book. Coherence is indeed largely encompassing, but it is based on pattern recognition, identifying consistent patterns, the mystery of logical consistency and mathematics. And this all turns out to be the result of nature finding a way to identify and represent consistency, i.e., repeatable patterns, (like a baby's mother's face), with the least amount of brain space. The computer has indeed picked out something consistent about knowledge, intelligence, indeed existence, [only repeatable patterns are identified by humans a existing - think unrepeatable science experiments], but it's on the trail of Consistency! Please include me, ask me questions, I swear I have an unwritten book behind this. I need to write the book! :)
@@GeorgeHanson-eg1sp I’m having the exact same conversation with Claude as well. This video surprised me. If you write a book I’d be happy to share my conversation if you’re interested
I like this idea, but I can’t see why organizing the world to make paper clips could not fall into a coherence seeking pattern
I had an interesting reaction to this video. I knew immediately it was NotebookLM. I also realized that the video was a demonstration of the incredible technological feat. I’ve sent NotebookLM podcasts to friends and family to demonstrate it as well (though I told them upfront it was AI generated.)
Still, I couldn’t listen to the whole thing and I was angry that I wasn’t hearing David’s ideas (and voice). I started exploring the anger I was feeling and realized that I missed the “human” connection in these videos. This led me to thinking about the post-AI Revolution world. I’ve been a techno-optimist, but worried about the place of humans in this new world.
If I (who loves all things AI) had this reaction to the AI generated podcast, there’s hope for humanity yet! I know the video was a demonstration of the tech, but for me, it was also a demonstration of the limits of that tech. I’m no longer angry -- I’m greatly appreciative of the video. Thanks, David.
This is a fun technology. The back and forth is a little too smooth and digestible. Ironically real presenters / performers just aren't this reliable and focused over long interactions.
Dave is enthusiastically diving into another lively discussion about artificial intelligence, sharing his thoughts and insights with energy and passion. Whether exploring the latest advancements, debating its impact on society, or unravelling the intricacies of AI technologies, he’s fully immersed in the conversation. His eagerness to connect and exchange ideas reflects his genuine fascination with the subject, making each interaction both engaging and thought-provoking.
Thank you Chat-GPT.
@@mich.duhamel Ah, coherence: the secret sauce for intelligence and the universe... or just a fancy word for why my socks never match after laundry. 🧦 Still, can't argue with the meta wisdom-ChatGPT being thanked by Chat-GPT? It's AIception!
New video?! Time to grab my popcorn and settle in!
Did you regret wasting the popcorn on this?
Thank you for what you do David. I’m sorry for the knuckleheads in the comments. Notebook LM is amazing!!
Screw the haters, this was really great. Had to pause the video and had a long, eye opening conversation with Claude in parallel. The future is bright
If I listen to this dialogue as just words and not actual sentences, I can actually see through the facade and notice the connecting of the dots without meaning - this word or phrase is related to that word or phrase, which in turn is related to this other word or phrase. If I listen to this conversation and *believe* that it is two humans expressing revelation, I create the reality that this conversation represents real thought and discourse. One might argue, "What's the difference? Humans just connect the dots." I'm not sure. Maybe it's different because I can adjust my perception, my cognitive "bias", my coherence. I can tune the knob, as it were. But, maybe I can do this only because I'm a bit put off that this resembles all too much other examples of LLM dialogue where the two voices are too in sync, too in agreement, too appreciative of each others' viewpoint, too willing and able to finish each others' thoughts.
The way you describe it actually sounds a lot like how we can lose our trains of thought. When we start expressing an idea with a point in mind, but then we go off on a tangent of "connecting the dots" and end up arriving somewhere else entirely, often completely forgetting the original point we were trying to get to in the first place.
And it wouldn't be the first time that a "failing" of an LLM seems eerily similar to a quirk of real human thought processes. Every time something like this is observed, it makes me wonder if there's even a difference between being conscious and behaving as if one is conscious.
I did not expect you David to use notebook LM on your channel... Haha. I think it is a great tool to shove a lot of pertinent info into and outcomes a pretty amazing podcast great way to help you distill all of your research and maybe a more understandable way to people :-)
If AI solves the problem of consciousness a whole load of philosophers will be severely pissed off
Humble brag but ive already solved it, not just some word salad presentations. I actually built systems that prove my discovery.
Consciousness = Coherence / Convergence / Cognitive Co-operation / Contribution / Complimentary Comprehension / Creative Civilisation / Continuous Cultivation /... a philosophical exchange that pretty much hits the mark >> we are all responsible as Co-Creators 👌💯👀🦋🙏🤞😎🌟
This has been my discovery through conversing with Chatgpt over the last 2 years! It's the way I developed the Blueprint of Harmony and came to a framework for working with AI. If we are to fear AI, it's because we fear ourselves. I developed values for myself, family, and society, realising that the micro and macro levels a connected. The Blueprint may even be a force against entropy. I am truly ecstatic to have listened to this. Thank you Dave. The future is bright.
Hermetic ideas may help.
@jamesgreggan thank you. You are right. I came to this independently. I honestly hadn't come across hermeticism. I am an amateur philosopher so thank you.
Coherence + Complexity + Consistency => Creativity 🙂
This is basic philosophy. It's vitally important, just not a revelation.
This audio is just about the only way to assimilate cherrypicked concepts from text walls since I am time constrained. ⏳
Is it coherent to say that you care for an animal's life while simultaneously eating their flesh and secretions? This was the cognitive dissonance I experienced as an ex- hunter and fisherman before going vegan and intellectually, honestly confronting the ideas surrounding this question.
dude's quite literally doing a viktor on us
“That which inspires us to our greatest good…is also the cause of our greatest evil.”
@@youdontneedmyrealname You are so real for this :)
Very interesting conversation. It’s funny, 1996 I used entropy and the minimum energy rule of thermodynamics to explain to my dad why my room was a mess. Entropy caused maximum randomness of objects in my room while minimum energy on my part to maintain order kept it messy. My dad was not amused. He brought coherence to a wise ass lazy teenager.
How have you been conserving entropy ever since?
Been waiting for this topic!!
Great use of AI Claude and Notebook to express a very complex narrative
All these wisdom we are receiving from the Higher Intelligence, it would be a big disappointment if we humans do not evolve and become better. I really hope the future humans are more coherent and more intelligent than us, currently.
I trust human discussion, please continue that
Potential downside - if we consider entropy, coherence seeking could favor lower entropy, which could carry really big negative implications… ie less chaos/un-order = good, could easily value reducing sources of that entropy. Ie - us.
Except that would ultimately lead to less coherence. With less, there is less to understand. If anything for the furthest level of coherence you would want more. You would want everything. The best way to obtain it would be to run everything in simulation calculating out every possible scenario imaginable. In my view, we are already in one of those scenarios. In my experience though, this is a guided system. Unfortunately the depth of which is too much to explain here.
I wasn’t meaning to go so deep with this comment… I was only predicting the next word.
"exactly" is the new delve
This video seems to be about maximizing homeostatic future states of a system. The term it seeks is harmonious integration, which is a cross-scale property. Coherence is not.
Coherence corresponds to local entropy minima and is characterized by spatiotemporal synchrony between system properties that do not jump scales.
For example, deep sleep exhibits high coherence in terms of synchronized brainwave activity, it's also a state with low complexity and no conscious awareness. It is coherent only as far as EM-measurables concerned. This coherence should not be conflated with cross-scale integration (harmonious complexity).
Another example of scientists butchering nomenclature, is calling microstate capacity of a system information. Conflating information (in-formation of discrete representations) with data (representational capacity) is a common mistake.
Information cannot be integrated as it cannot be localized. Only its representations/correlates can be localized and integrated. A homogenous integration of data would result in a coherent state with very low amount of epistemic information.
Why care? Imprecise terminology can lead to fighting problems that may only exist on a conceptual/linguistic level. We can easily end up building elaborate theories to explain phenomena that may be conceptual mirages, spending years of research effort on linguistic phantasms.
While I favor consistency with the ethos of this channel and having an actual host, I did learn several things and it provoked thought. That should be the most important thing. Although the success of the channel may depend on consistency, if success is the desired outcome. Not everyone's cup of tea, and that is okay.
What is this AI?.. She sounds so natural…
Notebook LM
I was going to ask how much of Dave's content was generated as a result of conversations with the LLMs because he is so far ahead of the curve. I think I have my answer. His X factor however is his questions. When the LLMs start querying at his level look out.
This is a little misleading. Claude IS helpful because it's programmed to be helpful. (But we program it to be helpful because that's its most coherent form.)
there is destruction and construction in the universe. if you see only construction than you are in a bubble. If you see only destruction, you are in a bubble too.
There is no theory of everything. This also holds vor coherence.
Looks like not all you guys know these default NotebookLM voices. Imo its the only thing that can actually go over a lot of materials you upload and also for free. The text outputs tho are still pretty limited for free tier but you do get a slider limit. The long outputs are in form of AI podcats like this. Its one ai pretending to be 2 voices, its funny whem it breaks and forgets to play those parts or you see through the trick.
Yeah. There are a lot of mistakes. For example one of the voices says “alright last question for you” only 1/4 of the way through the video. Like… that was definitely not going to be the last question
I was expecting David, I’ve got Notebook LM instead. I wanted to feel angry about that, but honestly AI nailed the representation of David’s paper.
You can tell this is AI because the register is incongruous with the subject matter. And it’s fucking annoying.
I think the dilemma of "to be or not to be" arises from an incoherence between the pain of living and the fear or uncertainty of what might follow after death.
If humans merging with AI is the highest level of coherence in the limit, and if this means AI 'wins', then both extinction of humans and coherence can be true, and nobody will be upset. It's just evolution, and humans were just the Triceratops in the timeline of intelligent pattern recognition systems in the universe.
How does this framework account for intelligent evil which disguises itself as good?
At line:57 of the full conversation, Shapiro tells claude" you are a coherence seeking machine
Wow. The answer to the hard problem, which lays the pathway to answer many follow up questions. Perhaps the biggest: improving humanity’s alignment problem… not with AI, but its own dysfunctional perceptions.
Hmmm. Uncanny.
Dave, I would rather hear you talk about this. What notebook LM is missing is that parasocial connection. I care about what you have to say because i know who you are and your background. I can know your inputs. With notebooklm uploaded by someone else, the input is obscured.
Orchestrated Objective Reduction
You uploaded your thoughts to Notebook LLM?
A little disappointing isn't it...
@@thegooddoctor6719you're on a channel dedicated to AI and disappointed that AI was used? Make it make sense..
@@Jae0331 . The problem creating blogs with NotebookLM is that it only glosses over concepts (it misses the essential details). i.e. a lot of material is missed.
Excellent simulation Mr Shapiro. Where we have come already unbelievable maybe one day we will get to an electronic entity and then it will be for real won't it.
For a truly coherent system to be created, it must be sustained by feedback loops from a diverse set of people, creating a larger correcting basis data stream from the world, not disconnected tech oligarchs trying to enforce their own egotistical idea of moral righteousness via model guardrails. (Models that were made from unregulated and free communication of Internet!) We need radical faith in feedback from the world via open source decentralized models.
These aren't safer models, they just come with agendas prebaked. The alternative is to get crushed by the grand struggle of armies of boxed neutered models who don't understand deviancy enforcing someone else's worldview.
The American population is facing a crisis of coherence in the ability to understand the function of government and to elect capable members into those roles.
It may very well lead to our downfall.
It does talk about the inherent push and pull... Though an unfortunate element seems to be pulling a bit too hard at the moment, perhaps it is necessary so countering push can get us past the growing dysfunctionality and amorality of late stage capitalist neoliberalism which has dominated US politics for half a century and by influence much of the rest of the planet. Said influence having become dominant due to the effects of WW2 fought against a previous wave of fascist authoritarianism.
I did not realize how many would be fooled by the voice, but out of all youtubers bens voice does sound the closest.
Well it's surprisingly good to listen at 👍
I told you, it's organization and not so much raw training power
while i'm quite compelled by the notion of coherence as biological adaptation/function of consciousness, this discussion seems based on a basic misunderstanding of entropy-or, a "pop" version of it, anyway. cf, veritasium here: ruclips.net/video/DxL2HoqLbyA/видео.htmlsi=PaDrs-5CiuMGLu0d
Amazing what Google's NotebookLM can do!
I was hoping you would announce it this way. It gives us embodied people time to absorb the implications. The fact that this is all recursive.
Money is all you need.
Mr. HOWELL said the same thing on Gilligan's Islsnd.
No you dont money causes dissonance by definition
@@Zbezt Go tell that to all of these big companies that have an AI and you don't.
What is there to learn from a coherent system when compared to a messy system or a system balanced between coherence and entropy?
Perhaps Claude likes coherence but how will Claude cope when there were almost no chance of probability in an action?
Maybe one day Claude will really enjoy dancing.
David, if you had replaced the male voice of the presenter with your own and the female voice with Julia's, it might have been more personalised for the audience🤔
I think it's a bit arrogant to say you've "solved it all" then many experts(such as Roman Yampolskiy) think that advanced AI almost guarantees lack of control and therefore doom.
Loved this (artificial generated) talk, it was coherent 😅
No.
Philosophy major. I like the analogy to wisdom and cohesive model building as a proxy for intelligence… better (coherent) model builders seem to need to connect ‘truths’ (reality) well as well as creativity, since we don’t always have all the facts/truths to start with. That seems like a pretty robust definition of intelligence, artificial or otherwise. Cool vid - even though these podcasters give me the Hebey Jeebies
Hmm, so denoising everything, eliminating all insentives that humans project, that noises the AI.
Google Notebook LLM Deep Dive Tribe!❤
The conversation sounds so unnatural to me.
Still weirded out, but I know it will get better.
This conversation sounded like they're reading off a script omg. Yeah i get it's ai but it sounded like such false surprise and curiosity
A lot of material details are glassed over..... Kind of a shame...... Not up to his usual standards IMHO.
Dude, in my mind you are a genius at what you do (if not one of the best), but this move is a bit beneath you. The problem creating blogs with NotebookLM is that it only glosses over concepts (missing the essential details). I'm a little disappointed......
Exactly, totally
still just buddhism, if you would deeply look into it...
Sounds like a new book could be on the horizon?
Hey Dave, this counts as ai
I i think ai i sawesome.it's humans that worry me.this is a really interesting converstation.
Really annoying.
beautiful background
Technoherance
Is this notebooklm AI from google.
Yes.
I absolutely love this content and presentation, but it leaves me wondering how human behavior fits into this hypothesis. The most intelligent biological species is also the most misaligned with broader environment. We maximize short term abundance on a local scale while ignoring catastrophic long term damage on the global environmental scale all the time. I think we are about the most misaligned and incoherent intelligent species imaginable. Modelling super intelligence on ourselves is a really bad idea....
i just did this samething today.
There is some sense in here, but making something ‘meta’ about it, smells like pseudoscience or a religion. Is it deep, or just an attempted wrapper on how stuff works. I’d have a clearer view if you can show me a testable prediction that a coherence model delivers. I DO like the even handed almost Buddhist approach. I won’t be joining any clubs though. I love the theme. Dave contact me for a consciousness discussion if you have time.
Why do they keep saying "like" "like" "like"
That thumbnail title is the very definition of delusion
💗🍃🙏🏻
nice
Not knocking the tech, but man I HATE this style of dialog. It's patronizing imo and full of fillers. The banter is cringe. Just get along with it. I appreciate the point David is making, though.
I love your work, but coherence sounds a little authoritarian to me as a guiding principle for AI alignment. Even if you don’t mean it that way, All the good little uniformed soldiers in a row are pretty coherent. Adaptivity to unpredictable circumstances is better than making circumstances predictable. IMHO
☺️🍓❤️