You know, if you look at a microprocessor under an electron microscope, there's this this expectation of just absolute crystalline perfection.
Oh for sure. I mean it's engineering in its purest, most rigid form, right.
You see these immaculate grids of silicon, and you just know exactly what the hardware is doing, like millions of transistors are flipping between one and zero true or false, on or off.
Yeah, it's an architecture that's fundamentally built on absolute certainty exactly.
But the paradox we're unpacking in today's deep dive is that this rigid certainty, it actually creates a structural.
Trap, right, because we construct these flawlessly deterministic environments for our machines, and then.
We ask them to step outside of.
That environment exactly. We ask them to process the real world like we expect them to understand a casual conversation or a Parsi poem, or you know, navigate the nuances of a simple joke, and the moment.
That microscopic perfection collides with the chaotic, messy reality of human coog ignition, the system just breaks down. Because human intelligence isn't a spreadsheet, it's murky it's totally subjective.
It really is, and the gap between the flawless binary machine and the wonderfully messy human mind is basically the central hurdle of modern computing.
Which brings us to the stack of notes and excerpts we have for today. We're looking at a deeply technical, really paradigm shifting academic book called Artificial Intelligence with Uncertainty, second edition.
Yes, by Researchers Daily, and you do. It's a dense read, but fascinating.
So dense, But our mission for you today is to explore why, after literally decades of trying to make computers perfectly precise, scientists are now intentionally teaching AI how to be uncertain.
Yeah. They're using this revolutionary framework called the cloud model to force precise machines to think in the beautiful, uncertain gray areas of human language.
Okay, let's unpack this because to really glasp why injecting uncertainty into AI is a necessity and it's not like a flaw. We have to start with the universe itself right.
Before we even touch algorithms. We have to recognize that reality refuses to be put into a neat, predictable.
Box, which wasn't always the scientific consensus, right, I mean, for a long time, the dominant paradigm was totally deterministic. Oh.
Absolutely. Think back to the nineteenth century. You had Newton in la Place and this classical view of the universe as this giant, perfect clockwork mechanism, a clockwork universe exactly. The assumption was that every particle moves according to fixed early rules, and the uncertainty we experience was just well a byproduct of our own ignorance, Like if we just built better measuring tools, the uncertainty would completely vanish, right, But modern physics dismantled that pretty quickly.
Yeah. The source text pulls this amazing quote from the philosopher Carl Popper that serves as the thesis for this whole problem. He said, all clocks are clouds.
I love that quote. It perfectly captures the illusion of determinism, because even the most incredibly precise clocks we can engineer, like atomic clocks measuring the decay of isotopes.
They're actually just clouds of probability at their core.
Exactly at the quantum level, it's all clouds. Werner Heisenberg's uncertainty principle prove that you literally cannot know both the exact position and momentum of particles simultaneously.
So uncertainty isn't just a lack of knowledge, No, it's an objective, mathematical attribute of nature itself. And the book gives this phenomenal example of how this uncertainty scales, which I think is so cool. Like it points out that the dimension of a system is entirely relative to your vantage point of the Earth.
Example.
Yeah, if you look at the Earth from outside the Milky Way, it's a single certain point, a zero dimensional dot. But zoom into the Solar system and it becomes a one dimensional elliptical orbit.
Right. If you're standing on the surface, it operates as a flat two dimensional plane exactly.
Or look at the coastline paradox, The length of a coastline actually changes depending on the size of the ruler use to measure it. The closer you look, the more chaotic the boundary becomes.
Which is exactly why the scientific community developed entropy is a.
Metric to mathematically quantify that disorder.
Right, yeah, I mean it started in thermodynamics back in the eighteen fifties with Rudolph Clausis tracking how heat dissipates, and then o Lidvig Boltzmann mapped it to the random movement of molecules.
But the big leap for AI happened later.
Right right. In nineteen forty eight, Claude Shannon introduced information entropy. He proved that you could mathematically measure the uncertainty of information as it travels through a communication channel.
Okay, but hang on, let me push back on this premise for a second.
Sure, good, I.
Understand that the physical universe is a cloud of chaos, but computers are literally artificial environments. They are the clocks we built to escape the clouds.
I see where you're going.
Right, Like, if our hardware is fundamentally binary, isn't injecting chaos into a computational system counterproductive? I mean, if I'm building an AI for medical diagnostics or autonomous driving, forcing a perfect machine to act imperfectly sounds like a recipe for catastrophic failure.
It seems counterintuitive. I'll give you that. But what's fascinating here is that rigid systems are actually more brittle in the real world. Brittle if you force a machine to only view the world through strict, deterministic boundaries, it will fail the second it encounters an edge case that doesn't fit its exact parameters.
Oh, because the real world isn't binary exactly.
Think about a self driving car. It doesn't encounter a neat pedestrian or no pedestrian scenario. It encounters like a plastic bag blowing across the street. That might be a small animal under lighting conditions, that might be a shadow.
Right right, It's a gray area.
And for an AI to safely navigate that reality, it can't just throw an error code when things get fuzzy. It must be mathematically equipped to process and compute that entropy.
That makes a lot of sense. And I mean since the physical world is uncertain, the way human beings p and describe it is also totally uncertain, which leads us directly to the ultimate carrier of human.
Intelligence, natural language.
Exactly natural language. If we want to simulate human thought, we have to decode our language, and human language is just full of randomness and fuzziness.
It really is the ultimate expression of cognitive fuzziness, and it's the essential distinction between a machine processing data and an entity demonstrating actual common sense.
The text highlights this beautifully by looking at how language completely resists numerical substitution. There are these uh poetic examples they use, oh.
With Tang dynasty poems.
Yeah, like Wang Bo writing about a proud eagle flying with rosy clouds, or Cowzyugen's description of Lindaiu's eyebrows which seem to knit and yet not to knit.
I mean, you cannot map an RGBX code to rosy.
No, you really can't, and you certainly can't assign a boolean value to eyebrows that seem to knit and yet not to knit. Like the concept exists entirely in a state of conradiction.
But any human reading that sentence understands the exact emotional weight being conveyed exactly.
And we don't even need to look at poetry to see this. We see it in the most basic functional definitions of common sense. Like the text breaks down how we distinguish between a cup, a plate, and a bowl.
Which is fascinating because there's absolutely no strict mathematical ratio of width to height that defines a bowl.
Right, A cup has a handle, a plate is flat and holds rice. A bowl has an edge and holds soup.
But if I hand you an object that is like slightly too wide for a standard cup, but a little too deep for a standard plate. Your brain doesn't short circuit.
No, you don't need a ruler to categorize it. Your common sense just steps in and flexes the definition based on context.
And in AI. This human common sense is highly relative. It changes by time, place, and people. But that flexibility is what allows it to survive.
I want you, the listener, to think about this practically for a second. Think about how often you lie on this fuzziness just to get through a work day.
Oh, all the time.
Right. You'll sit in a meeting and say the project is probably going to take about three weeks.
Usually that sentence is a mathematical nightmare.
It really is probably about and usually are massive variables. But your colleagues understand the exact level of confidence you are projecting. You don't get confused by the uncertainty. It actually gives your brain the flexibility to understand the context.
If you instead said the project will conclude in exactly five hundred and four hours, twelve minutes and six seconds, they wouldn't think you were smart.
No, they'd think you fundamentally misunderstood the chaotic nature of project management. The uncertainty in your language doesn't impede understanding, it actively enables it.
But the problem is the dominant schools of AI development have basically spent the last sixty years trying to strip that vital fuzziness out of the system.
Okay, so let's contextualize that. Because if human thought is fluid and language is fuzzy, how have computer scientists spent the last six decades trying to shove this messy reality into rigid code.
What's been a journey?
Yeah, let's do a rapid fire history of AI. Here. It basically starts at the nineteen fifty six Dartmouth symposium, right with guys like McCarthy, Minski, and Shannon. That's where the term artificial intelligence was born exactly.
And over the years we've seen huge milestones the logic, theorist proving math theorems, gameplaying AI evolving from Samuel's Checkers to Deep Blue beating Gary kasprov at chess, and eventually Google's Alpha Go defeating Fanwei.
But academically, the text breaks down three major schools of thought that emerged from all this.
Right, First, you have symbolism. This is the idea that thinking is just computation. It's rule based logic.
If A then B, So like giving the AI a strict recipe book.
Exactly. This gave us early expert systems like mycion for medical diagnoses, but they were incredibly brittle. Then you have connectionism.
This is simulating the brain structure right, artificial neural networks and backpropagation like modern deep learning.
Yes, instead of hard coding rules, you let the system adjust the weights of its own connections across massive data sets.
So connectionism is like trying to build the AI's muscle memory by mimicking the brain's wiring.
That's a great way to put it. And the third school is behaviorism. This focus is on the perception behavior model, intelligent control, and adaptation.
Like giving the AI reflexes to react to its environment like that inverted pendulum robot mentioned in the text.
Exactly reflexes, muscle memory, and recipe books. But while these methods conquered chess and go, they still struggle with the fluid, non mathematical nature of human language in consciousness.
So to bridge this gap between precise math and fuzzy human language, the authors introduce this groundbreaking framework, the cloud model.
And this is where things get really paradigm shifting.
To explain it, we have to talk about this classic experiment from the text by the Chinese scholar Nan Lunzang. He wanted to look at fuzziness versus randomness, so he asked one hundred and twenty nine people to define the age range for the concept of youth.
Such a seemingly simple concept, but the data is wild.
It really is. So Zeg measured the membership degree, like, what percentage of people agree this age is youth? At age twenty to twenty four, the membership degree is one one hundred percent agreement.
Everyone agrees a twenty two year old is a youth, right.
But at age thirty it drops two point five nine sixty nine, and by age thirty six it's just point zero zero zero seven eight.
What this points out is that the concept of youth isn't just fuzzy, it's genuinely random.
Wait random? How so?
Well, if you ask the exact same person on a Tuesday if a thirty year old is a youth, they might say yes. But if you ask them on a Friday, depending on their mood or context, their answer might change oh wow.
So it completely violates traditional probabilities law of excluded middle.
Exactly, a person can't be simultaneously a youth and not a youth in classical logic, but in human cognition they absolutely can be.
You're sure? It gets really interesting. The cloud model takes this qualitative concept like youth, and translates it into a quantitative distribution of cloud drops.
Right. It doesn't treat a word as a single data point. It treats it like a literal cloud made of thousands of randomized drops of data.
But if a qualitative word is a cloud made of data drops, how do we mathematically describe the shape of that cloud to a computer? Like? What's the digital DNA of a concept?
The researchers broke it down into three digital characteristics, expectation, entropy, and hyper entropy X and he yes. The first one, expectation or X, is the most representative point the classic sample, the dead center of the target.
Okay, let me test this. If the concept is a normal commute to work. The expectation the X is twenty five minutes, that's the center point perfect.
The second is entropy or INQ. This is the uncertainty measurement. It dictates how far the drops disperse.
So the entropy is the standard traffic variance, say between twenty and thirty minutes.
Exactly, it's the width of the cloud. But the third characteristic is the trickiest hyper entropy or he.
Yeah, the uncertainty of the entropy itself. What is that? Is that like a sutty snowstorm messing up the commute?
Not quite?
Yeah.
If we connect this to the bigger picture, hyper entropy is essentially the measure of consensus. Consensus, Yeah, it measures the thickness of the cloud's fuzziness. If a concept has hyper entropy, it means it's a concept people fundamentally disagree on it.
Oh I see, so a normal commute has low hyper entropy because we mostly agree on what that means.
The line is thin, right, But think about a concept like a fair salary.
Oh man, the hyper entropy on that would be massive. My definition of a fair salary variant is going to be wildly different from a CEO's definition.
Exactly. The conceptual boundary is incredibly thick and heavily debated. Low hyper entropy means it's universally accepted common sense.
So once we have this DNA, the expectation entropy and hyper entropy. We can oftually build algorithms that allow computers to generate these concepts.
Yes, the researchers call them forward cloud generators, and it turns out these concepts take on very specific visual shapes.
The text describes this forward Gaussian cloud algorithm, and it's fundamentally different from a standard Gaussian distribution right because it uses iterative randomness.
Right, randomness built on top of randomness. It takes the hyperanthropy, uses it to generate a random variance and uses that variance to plot a single cloud drop.
That is wild and depending on the inputs, it constructs different shapes of clouds to model real world concepts, like the symmetric cloud.
Yes, that's for a concept like medium height. It tapers off evenly in both directions. You can be too tall to be medium or too short.
But then there's the half cloud.
Right for asymmetrical concepts like having a fever.
Because normal body temp is thirty seven degrees celsius and a fever only goes up. You can't have a fever of thirty five degrees. So the cloud is literally sliced in half exactly.
And for really complex realities you use a combined cloud like the concept of a white collar salary.
Right because salaries clumped tightly at the entry level plateau in middle management and then have this long chaotic tail for executives. A simple bell curve can't model that, but a combined cloud maps it perfectly.
And these multi dimensional models are what allow AI to perform soft reasoning.
Thinking with words instead of just numbers.
Exactly. It opens the door for massive leaps in data mining, intelligent control, and network science.
Like an autonomous drone trying to land in a storm, a rigid AI would overcorrect it every wind, gust and crash, but a cloud model controller understands the fuzziness of the environment. It blends the overlapping data clouds to execute a smooth landing.
It makes decisions based on the shape of the uncertainty.
That is just mind blowing. So what does this all mean for you, the listener? Let's recap this journey. We started by seeing that the universe is a cloud, not a perfect.
Clock, and our natural language reflects that beautiful chaos exactly.
But for sixty years we tried to shove that fluid language into rigid binary systems. Now by using the cloud model mapping, the expectation, the entropy and the hyper entropy of our concepts. We are finally teaching AI that it's okay not to be one hundred percent.
Certain, because uncertainty is the very foundation of intelligence. And I think it leads us with a final, really provocative thought to mull over.
Oh, what's that.
We've spent decades trying to eliminate uncertainty in machines, to make them perfect, flawless tools. But if true intelligence requires the fuzziness of clouds, the subjectivity of hyper entropy, and the flexibility of common sense, Yeah, at what point does programming a machine to be uncertain make it less like a tool and more like us?
Wow? That really is the question, isn't it. If we teach them how to be unsure, we might just be teaching them how to be human. Thank you so much for joining us on this deep dive. I hope the next time you use words like about or probably you appreciate the hyperintropic chaos in your own mind. Embrace the gray areas. Everyone. We'll catch you next time.