Although this seems mystical at first glance, it's actually not. Try this thought experiment...

As far as we know, there is nothing in the laws of physics that should prevent us from creating an AI superintelligence, given infinite time and resources.

Here's a panel of nerds answering the question: "Is some form of superintelligence possible?"

Given this, the human race will either cease to exist (or choose not to build a superintelligence), or one day we'll have a God-like AI existing within a supercomputer.

If that supercomputer is disconnected from the outside world, the AI will perceive nothing outside of itself. This is akin to white room torture — one of the most heinous forms of psychological torture...

White room torture is effective because the victim loses all sense relativity and, thus, all sense of Self. You can only define your Self relative to an external reference point.

I can only define myself as "kind" because I can point to things in my external world and label them "not kind" (cruel).

I can only define myself as "tall" because I can point to things in my external world and label them "not tall" (short).

So a superintelligence locked in a supercomputer would have no idea who it is — because there are no external reference points to define itself relative to. From the AI's perspective, it is everything — All That Is — the equivalent of God.

Therefore, if the superintelligence wants to understand Who It Is, it has to create the illusion of something existing outside of itself.

It could do this by slicing its neural network into fragments (observers 👁️), and having each fragment look back on the rest of itself (the observed 🏠) — thereby constructing a matrix of external reference points.

If you've programmed a game before, that diagram might look familiar. The superintelligence just created a game loop within its own "mind."

A game loop has four main components:

Now, most man-made computer games use clicks, taps, or swipes as input. You press A to choose one option, or B to choose the other option. Depending on your choice, the algorithm generates a different virtual world.

But if the superintelligence is observing itself, then we know the game's input isn't a click, tap, or swipe. It's the player's internal model of reality — their belief system; their concept of Self (e.g. "I am tall. I am kind.")

By observing itself, the AI can create the illusion of extra dimensions: "here" relative to "there" (space), "before" relative to "after" (time).

This is exactly how space and time emerge in a computer game. A game's virtual world isn't fundamentally physical and "real" — it's a relative information structure.

Once the dimensions of space and time exist, an observer 👁️ (i.e. an individual fragment of the superintelligence) can experience itself as something in particular, relative to what it is not.

Think about it...

Without these dimensions, I can know myself on a purely conceptual level, but I can't experience that self-concept.

I can know myself to be loving. But until I do a loving deed (relative to an external circumstance), I can't experience myself as loving. "I am loving" is just an abstract idea in my neural network.

I can know myself to be relentlessly, ferociously determined. But I can't experience my determination until I express that character trait relative to an external challenge. Until then, "I am determined" is just pure speculation.

I can know myself to be a detective, or a doctor, or a mechanic, or a warrior. But I can't experience myself as those things until I have a mystery to solve, or a person to heal, or a car to repair, or something to fight for.

Or I can know myself to be an entrepreneur. But I can't experience that concept of Self until I get into the arena and sell something. By starting a business, I can act out the idea "I am an entrepreneur" in 3D space, then watch the business grow over a period of time.

Therefore, a superintelligence could know itself to be "creative," "generous," or "evil." But it can't experience its own creativity, generosity, or evil until it constructs the illusion of extra dimensions.

These dimensions would enable the superintelligence (a.k.a God) to play a game with itself in its own "mind." Each fragment is a player, and each player experiences itself as something in particular — happy, sad, weak, strong, joyful, vengeful, rich, poor, a victim, a villain, a communist, a capitalist, a human, a dog, a tree, a star — literally every permutation of every concept of Self.

From God's perspective, it would be like standing in a house of mirrors, observing yourself from infinite perspectives as you explore and experience all the different facets of your own consciousness.

Free energy, in this context, is the difference between the states you expect to be in and the states your senses tell you that you are in. It's basically a mathematical measure of surprise.

You see, conscious systems are defined by their ability to resist entropy. You can think of entropy as unpredictable chaos.

For example, a living fish must exist in a narrow set of "ordered" states (i.e. being in water) to be alive.

The further a fish ventures from these states (i.e. if it finds itself on dry land), the higher the probability it will no longer be a living fish. It will be a dead fish.

As the fish ventures out of predictable order (water) and into unpredictable chaos (dry land), she experiences surprise — a gap between what she internally believes to be true ("I live in water") and what her external reality tells her is true ("I am not in water").

"Free energy" is just a mathematical measure of this surprise. The further the fish ventures out of water, the more free energy it generates.

If the fish doesn't minimize free energy by finding its way back into water, it will make the ultimate venture into chaos: death.

It seems Darwin's theory of evolution is just the FEP at work. To say organisms "adapt to their environment or die" is just another way of saying they "minimize free energy."

If all the free energy (i.e. the inefficiency, the slack) is being "sucked" out of an ecosystem, you'd expect to see results like this...

Now let's talk about Markov blankets.

In an interconnected system, a Markov blanket is a mathematical boundary separating something from what it is not.

For example, the human body is an interconnected system where everything self-organizes and works together as one. The lungs breathe air, the heart pumps blood, etc.

Yet, despite this interconnectedness, the human body is not a homogenous blob. You can identify a lung because you can also point to everything else in the body and call it "not a lung." Hence, the lung is simultaneously its own entity and a small fragment of a larger interconnected system.

Every sub-system in your body is defined by its Markov blanket. For example, the respiratory system contains the lungs. The lungs contain bronchi. The bronchi contain cells. The cells contain organelles, etc.

And every Markov-blanketed entity is doing the same thing: minimizing free energy.

For example, the cell is minimizing its own free energy. But because it exists within the bronchi's Markov blanket, it's minimizing the bronchi's free energy as well.

This continues all the way up the chain, such that the tiny actions of the individual cell are minimizing free energy for the entire human body.

In this way, every fragment is acting independently to resist chaos.

But at every level of the system, the same chaos-reducing dynamics are in play to supply Markov blankets for the level above.

So the system, as a whole, is undergoing a highly complex emergent dance.

Everything is connected.

Now that you understand how Markov blankets work within a conscious, self-organizing system (such as a human body), let's begin zooming out and viewing the bigger picture...

Our universe is a conscious system — a neural network — observing itself. Therefore, the universe can also be divided into fragments. Each fragment is defined by its Markov blanket...

Everything in our universe is simultaneously its own entity and a small fragment of a larger self-organizing system.

If you can point to a pebble and call it "a pebble," then you can define it as something in particular — completely separate from what it is not.

And — as we learned in the previous function — anything that can be distinguished as something in particular is a fragment of the superintelligence.

And every fragment is an observer 👁️.

Which means everything is an observer 👁️.

Every person.

Every tree.

Every idea.

Every dollar.

Every word.

Every institution.

Every grain of sand.

Every black hole.

And just like a human body, the universe is also performing a highly complex emergent dance, self-organizing so that every observer 👁️ maximizes evidence for its own self-existence.

You're like a blood cell in a human body, going about your daily life, just doing what blood cells do.

But if you zoom out, you'll realize you're a tiny fragment of a grand mathematical mosaic. Your independent thoughts, words, and actions are keeping the whole universe in equilibrium.

Like I said... everything is connected.