The GigaFactory story @17:28 reminded me how powerful NVIDIA's Omniverse Digital Twin product can be in designing and maintaining a production facility.
That thing… I’m sure they’ve used smtg similar to build their new GPU architecture (where they’ve multiplied the processing power by 18 in only one « real life » iteration.
These are all great! a couple points: the zeroeth principle thinking model is excellent for reasoning, but when applied to life, we need principles, like morals. With the distillation model, I use this all the time, but we have to remember that when we compress information, important details can be lost
I prefer Gen Alpha's take on postmodernism. When they see a paragraph more than three sentences, they reply I AINT READIN ALL DAT 🗣️🗣️🗣️🔥🔥🔥🔥💀💀💀 truly befuddles all minds (they have no theory of mind)
True. On debates no party will learn anything. The most important thing of a debate though, is to inform the people watching. Hopefully who watches will get a better more nuanced idea instead of falling into dumb ideology.
I'm almost through "The Status Game" and it's easily one of the most enlightening books on human nature that I've read. Highly recommend David's reading suggestion!
Hey love you channel. Do you think the FEP and active inference is ultimately a better way to achieve genuine AGI. Karl friston is critical of LLMs. Basing AGI on how nature works seems to be a better approach. I wondered whether you could do a video comparing Karl Firston’s approach to AI vs LLM. The FEP also how some metaphysical assumptions about the nature of consciousness which are controversial within some circles but it does tie nicely with non dual spiritual traditions…
I love the "Okay, now get out of that ivory tower and make your hands dirty in the real brick and mortar world." take on post modernism. Thoughts that are so nihilistic that they actually prevent people from thinking are not valuable. Useful thoughts eventually create understanding of the real world. Also, if I might add, with destination its is also good to think of what the generalized definition excludes. And if it is not possible to come up with a good precise definition, maybe the thing perceived is not a real category. Like the concept of a (biological) tree for example.
The Socratic Method is why I love Eliezer even though I am his "adversary" and I think 99% of his arguments are science fiction. We need a wallbreaker for each wallfacer ya know? Makes em both up their game. I can tell you spent years in "IT".
The slug is also built up in a weave of predictions, FEP happens from its nervous system all the way down to its individual cells. Like Friston & Michael Levin are saying, the human doesn’t have any special principle underlying its behavior, its just manifesting more levels & a greater degree of feedback
@@Systems.Thinking Folks like Michael Levin or Shamil Chandaria keep saying that there’s nothing special about the brain. There’s cognition going on in other organs, tissues, cells & unicellular organisms, there’s prediction & free energy minimization even in the simplest bacteria. You can definitely build novel abilities along with complexity but there’s no unique principle instantiated in humans. Solutions evolved for regulating homeostasis in chemical space carried across to organizing bodies in morphological space and then to model movement in 3D space, and of course, things used for organizing movement in 3D space got recruited to organizing our thoughts in the space of abstract representation. Michael Levin’s famous words, change neural tissue with any other cellular networks & change milliseconds with longer timeframes, and you basically have the same underlying principles at work and you can use the same neurological theories & methodologies to study behavior. Also, like others are saying, the intuitive "System 1" *isn’t* different at heart to the logical "System 2". The latter is a sort of virtual machine runnin on the former. Or we could say that it’s a set of constraints imposed on it making for a more formal rigorous dynamic. "We don’t have a native instruction set for rationality. We’re intuition machines that trained on logic", Carlos E Perez talking about it on Twitter.
Yeah... no. If there was nothing special about the human brain then we would have a world full of super intelligent animals. Yes, the body is intelligent in its own right but that utterly fails to recognize the power of language, abstract thought and systems thinking.
Probably this comment will be lost to the ether but...I post because I did watch and thought I might offer a few thoughts... The first is that I have been leveraging "Systems Theory" since the first time I read von Bertalanffy...There are many great books (and a strong theoretical backbone) but not a useful curriculum for a long time... I wonder what your commentary is regarding actual physics and "systems theory". Examples would include: - Robert Rosen ("Life Itself") and similar physicists - Prigogine (see above) - More recent work including Wolfram and V.N. Pokrovskii...where scientists are actively acknowledging current limitations as well as proposing useful frameworks to address them... thanks if you read this and actually answer!
Great video. Would love more Zero System’s thinking approaches for technical problem solving, in particular for rooting out those early assumptions made in a pre-Superintelligence era that should be challenged to achieve a better solution
Here is a distilled definition of intelligence that I really like: Intelligence is the ability to set and achieve a goal. I really like it, and it has been very practical to me. It allows me to see where I've been intelligent and where not, and also helps me choose the (expected) best course of action. I also think it's really useful for gauging the intelligence of AI. And by understanding that reaching a particular goal depends not only on the AI, but on how we comunicate with it, it becomes obvious that the total intelligence emerges from the interaction of both parts.
@@Systems.Thinking It gets deeper the more you think about it: The harder a goal is, the more intelligent you need to be. It doesn't take much intelligence to make 1$ per month. 100k maybe takes more. If you own a huge company, maybe it's not that hard. If your starting situation is more challenging, like being born in a 3rd world country with no parents, then it requires more. Time also is a factor. Solving a differential equation doesn't require much intelligence if you've been studying them for ages. If you have no previous training in math and you solve it, you surely are intelligent. If on top of that you do it in under 5 minutes, you are probably a genius. It also implies that intelligence is without exception attached to facts. If you can't solve a simple arithmetic problem now, you are not intelligent enough to reach this goal now, no matter how many you solved before. Intelligence doesn't exist in a vacuum, it requires some form of interaction with reality. Setting a goal is also part of being intelligent. That word is there for a reason. Autonomy implies higher intelligence, but also setting goals that you can achieve. All this comes from the same distilled definition. It leads to many avenues that you can explore, these are only some.
The GigaFactory story @17:28 reminded me how powerful NVIDIA's Omniverse Digital Twin product can be in designing and maintaining a production facility.
That thing… I’m sure they’ve used smtg similar to build their new GPU architecture (where they’ve multiplied the processing power by 18 in only one « real life » iteration.
I love how every hypothetical or actual being of any kind prefaces every thought or statement with "hey". (Non-sarcastically & unironicially).
These are all great! a couple points: the zeroeth principle thinking model is excellent for reasoning, but when applied to life, we need principles, like morals. With the distillation model, I use this all the time, but we have to remember that when we compress information, important details can be lost
I prefer Gen Alpha's take on postmodernism. When they see a paragraph more than three sentences, they reply I AINT READIN ALL DAT 🗣️🗣️🗣️🔥🔥🔥🔥💀💀💀
truly befuddles all minds (they have no theory of mind)
Love what you do Dave, your channels are underrated ❤
True. On debates no party will learn anything. The most important thing of a debate though, is to inform the people watching. Hopefully who watches will get a better more nuanced idea instead of falling into dumb ideology.
Great work 👏. I just realized how I have applied systems thinking in my clinical hypnosis practice. Thanks for the Illumination.
I'm almost through "The Status Game" and it's easily one of the most enlightening books on human nature that I've read. Highly recommend David's reading suggestion!
Brilliant! I was actually looking for understanding those principles! Thank you!
Cognitive Covergence FTW 🙏
Hey love you channel. Do you think the FEP and active inference is ultimately a better way to achieve genuine AGI. Karl friston is critical of LLMs. Basing AGI on how nature works seems to be a better approach. I wondered whether you could do a video comparing Karl Firston’s approach to AI vs LLM. The FEP also how some metaphysical assumptions about the nature of consciousness which are controversial within some circles but it does tie nicely with non dual spiritual traditions…
Transformers are the way. People forget that LLM isn't an architecture
I love the "Okay, now get out of that ivory tower and make your hands dirty in the real brick and mortar world." take on post modernism.
Thoughts that are so nihilistic that they actually prevent people from thinking are not valuable. Useful thoughts eventually create understanding of the real world.
Also, if I might add, with destination its is also good to think of what the generalized definition excludes. And if it is not possible to come up with a good precise definition, maybe the thing perceived is not a real category. Like the concept of a (biological) tree for example.
My brain is a Cyber-Truck 😎
David, top 5 podcasts you would recommend?
KM Weiland and Writing Excuses.
The Socratic Method is why I love Eliezer even though I am his "adversary" and I think 99% of his arguments are science fiction. We need a wallbreaker for each wallfacer ya know? Makes em both up their game.
I can tell you spent years in "IT".
The slug is also built up in a weave of predictions, FEP happens from its nervous system all the way down to its individual cells. Like Friston & Michael Levin are saying, the human doesn’t have any special principle underlying its behavior, its just manifesting more levels & a greater degree of feedback
I don't fully agree with that. The human brain presents an abstraction layer.
@@Systems.Thinking Folks like Michael Levin or Shamil Chandaria keep saying that there’s nothing special about the brain. There’s cognition going on in other organs, tissues, cells & unicellular organisms, there’s prediction & free energy minimization even in the simplest bacteria. You can definitely build novel abilities along with complexity but there’s no unique principle instantiated in humans. Solutions evolved for regulating homeostasis in chemical space carried across to organizing bodies in morphological space and then to model movement in 3D space, and of course, things used for organizing movement in 3D space got recruited to organizing our thoughts in the space of abstract representation. Michael Levin’s famous words, change neural tissue with any other cellular networks & change milliseconds with longer timeframes, and you basically have the same underlying principles at work and you can use the same neurological theories & methodologies to study behavior. Also, like others are saying, the intuitive "System 1" *isn’t* different at heart to the logical "System 2". The latter is a sort of virtual machine runnin on the former. Or we could say that it’s a set of constraints imposed on it making for a more formal rigorous dynamic. "We don’t have a native instruction set for rationality. We’re intuition machines that trained on logic", Carlos E Perez talking about it on Twitter.
Yeah... no. If there was nothing special about the human brain then we would have a world full of super intelligent animals. Yes, the body is intelligent in its own right but that utterly fails to recognize the power of language, abstract thought and systems thinking.
Probably this comment will be lost to the ether but...I post because I did watch and thought I might offer a few thoughts...
The first is that I have been leveraging "Systems Theory" since the first time I read von Bertalanffy...There are many great books (and a strong theoretical backbone) but not a useful curriculum for a long time...
I wonder what your commentary is regarding actual physics and "systems theory". Examples would include:
- Robert Rosen ("Life Itself") and similar physicists
- Prigogine (see above)
- More recent work including Wolfram and V.N. Pokrovskii...where scientists are actively acknowledging current limitations as well as proposing useful frameworks to address them...
thanks if you read this and actually answer!
My math degree might actual come in handy finally…..
Great video. Would love more Zero System’s thinking approaches for technical problem solving, in particular for rooting out those early assumptions made in a pre-Superintelligence era that should be challenged to achieve a better solution
Have you discovered Street Epistemology yet?
The podcast is Bryan Johnson and Tom Bilyeu lol I listened to it also
That was one of them yeah, there were a few others
35:57 - "someone once called me a delaton not a polymath"
What's "delaton" or whatever the correct spelling is? what does it mean?
Dilettante - Somebody who dabbles in an area out of casual interest and not professionally.
@@benburris4735 Thank you!
I miss the old profile picture. It was based.
Different channel.
@@Systems.Thinking haha just noticed
Here is a distilled definition of intelligence that I really like:
Intelligence is the ability to set and achieve a goal.
I really like it, and it has been very practical to me. It allows me to see where I've been intelligent and where not, and also helps me choose the (expected) best course of action.
I also think it's really useful for gauging the intelligence of AI.
And by understanding that reaching a particular goal depends not only on the AI, but on how we comunicate with it, it becomes obvious that the total intelligence emerges from the interaction of both parts.
I like that definition. Very distilled. I might just have to adopt that.
@@Systems.Thinking
It gets deeper the more you think about it:
The harder a goal is, the more intelligent you need to be.
It doesn't take much intelligence to make 1$ per month. 100k maybe takes more.
If you own a huge company, maybe it's not that hard. If your starting situation is more challenging, like being born in a 3rd world country with no parents, then it requires more.
Time also is a factor. Solving a differential equation doesn't require much intelligence if you've been studying them for ages.
If you have no previous training in math and you solve it, you surely are intelligent. If on top of that you do it in under 5 minutes, you are probably a genius.
It also implies that intelligence is without exception attached to facts.
If you can't solve a simple arithmetic problem now, you are not intelligent enough to reach this goal now, no matter how many you solved before.
Intelligence doesn't exist in a vacuum, it requires some form of interaction with reality.
Setting a goal is also part of being intelligent. That word is there for a reason. Autonomy implies higher intelligence, but also setting goals that you can achieve.
All this comes from the same distilled definition. It leads to many avenues that you can explore, these are only some.
"Respect the subjective experience of others" is a good one too
¹1¹st;
I'll drink to that !¡!
4:49
🍺
.