Elias Ruokanen

INTRO

 

AI hype has reached astronomical proportions. It is not enough now to state that artificial intelligence is a technology that will reshape society. It is something more. The term "artificial intelligence" radiates immediate, un-nameable meaning - a substance into which people project divine salvation on one hand and The of the World on the other.

 

This fearful confusion and amazement at the feet of the unknown tells us that we are in the realm of spirituality. The downside of this situation is that as long as it is not recognized, it is difficult to have a conversation.

 

One feels forced to relate to AI with a deeply serious, hush-toned reverance, as if one were in the presence of a supernatural entity. At the same time recognizing the mystery of the human mind is felt to be naïve and petty, like desperate nostalgia for outdated furniture that is about to be kicked to the curb.

 

This state of affairs is not only questionable, but annoying. People appear to be perfectly willing to debase themselves in order to project something more onto a computer program that people designed.

 

 

GIZMO

 

I think AI is just a tool, were it doing statistical analysis or "mapping out a strictly defined solution space." But if I want to use this tool I am pressured into accepting a whole ideology. This is kinda weird. If I buy a power drill, I am not inculcated into the power drill cult. Power drill manufacturers only want to sell power drills. 

 

Programs/apps ship with a compulsory philosophical add-on that is hard to turn off (like many things in Windows). The Silicon Valley view of human beings spreads through our culture like a parasite, impelling people to change their identity to a form more compatible with computer language.

 

You can sense it in programs that assume to know what you want to say better than you do - and write your messages for you. You can observe it social media platforms that herd people into certain behavioural paths, so CEO's can tell marketers how easy it is to read the human mind.

 

When fully incorporated into oneself this cultural gizmo inhibits a person from recognizing their own autonomy, and thus their own responsibility. When Elon Musk was asked¹ why he bought Twitter, he replied, "my biological neural net said it was important." A neural net is a type of AI.

 

The worldview being pushed to people like a forced software update is the following: A person is just a computer. From this follows: A computer can be a person. And from this follows: A computer can be a conscious being.

 

This worldview is presented as essentially complete, but it has a person-sized whole in it: Computers don't objectively exist.

 

At this point I request some patience from the reader.

 

 

GUESS WHO?

 

The Finnish word for computer is a fine word, tietokone. It literally means information (tieto) machine (kone). A computer is a machine that operates on information. Information is transmuted into a malleable form by turning it to bits, the most fundamental unit of information.

 

A bit is just the simplest answer to a question. Yes or No. A thing is or it isn't. In computer language we use the terms one and zero.

 

By chaining questions and answers we get more complicated information structures. If you have ever played the classic board game Guess Who?, you have realised your own inner computer scientist without knowing it. In the game players choose a character from an assortment of character cards, without showing it to their opponent. Then the players try to guess each other's character by asking yes-or-no questions.

 

Is he a man? No. Is she old? No. Does she have red hair? Yes. Does she have blue eyes? Yes. Does she have a lip piercing? No. You get the idea. This is continued until we reach the right character. We can create a chain of bits (e.g. 00110) that now represents that character.

At this point an interesting question arises. How do we know that our chain of bits represents that character? The same chain of yes-or-no answers could correspond to any other character in the game - it depends entirely on the questions. The same chain of bits may well represent the meaning of life, if we just ask the right questions.²

 

In other words the freestanding, independent existence of information is not so clear-cut. Does our chain of bits correspond to a character in a board game? In the words of an old Finnish sketch show lampooning the laconical cruelty of Eastern Finns: suattaapi olla, suattaappi olla olematta. It might, it might not. (In the sketch a blind woman is asking a man if her bus stops at this station.)

 

 

THE GATEWAY DOMINO EFFECT

 

Let's imagine a computer. We can build it out of domino blocks. To make our lives easier we install little motorized hinges onto the dominoes, so that after falling they bounce back upright.

 

By placing dominoes into different rows on the floor of our room we can construct the logical gates that make up a computer. A logical gate let's a signal through if its conditions are met.

 

For example, we can imagine a domino block that is fatter and heavier than the others. A regular old domino can't topple a fat domino. If we place this fat domino on some section of our row (call it row 1), the dominoes will stop falling at that spot. Observe the ASCII art demonstration:   / / / / / /┃┃┃┃┃

 

But if we assemble another row of dominoes (row 2), we can position it to strike the fat domino at the same time as row 1. Two regular old dominoes join their forces to topple the fat domino. Row 1 and 2 join up (like the lines of the letter Y) and the fat domino acts as a gate.

 

This is called an AND gate. Two dominoes (from rows 1 and 2) must fall on it at the same time, otherwise the signal stops. If we replace the fat domino with a regular one in this scenario we implement an OR gate. As long as it's hit by a domino (could be from row 1 or 2 or both) it falls and the signal propagates onward.

 

There are different kinds of logical gates, and by combining them we make the rules by which the dominoes fall or stand upright along our computer.

 

 

ONE MAN'S ALARM CLOCK IS ANOTHER MAN'S...

 

In the starting position all dominoes are standing at attention, and we have (let's say) 50 rows of them at the starting line, which then branch out and join up later in different ways, like streams joining to form a river.

 

The decisions to push or not push any one of the dominoes on the starting line represent the program's input. I knock down / I do not knock down, Yes / No, One / Zero. We can prod the dominoes ourselves, making us part of the machine in a a way³, or we can construct a prodding mechanism of some sort, so that the hardware purists among us remain satisfied.

 

The network of domino rows then finally reaches its end somewhere. They all converge onto a single domino, which if pushed, will strike an empty metal bucket.⁴ Now we have a machine that under certain parameters plays a *clang* sound.

 

Let's add one final mechanism, a temporal one: we make circular segments, loops, which dominoes repeat an X number of times, according to our input, before moving out of the loop and continuing on. Now we have an alarm clock!

 

We start our machine. The blocks are prodded and begin to fall in wonderful orchestrated dance. However, due to a cruel twist of fate, we have left a window open in our room. The wind blowing from it twists one of our dominoes from its place. It strays too far from its predecessor domino and cannot be knocked over.

 

Now the input that would have set the time for our alarm clock no longer functions as intended, and the *clang* is played at random times all throughout the day. Do we now have a broken computer: just a bunch of dominoes and a lonely bucket? Not necessarily! If we interpret the *clang* of the bucket as the beat of a metronome from a rhythmically exotic culture, we have a new and functioning computer.

 

The existence of a computer is dependent on the human context in which it appears. This is what the Faithful of Artificial Intelligence, who regard persons as chains of bits, don't seem to realize. We have no objective way of detecting a computer.

 

Jaron Lanier, the programmer and philosopher who coined the term Virtual Reality, put it thusly in a paper⁵ published over two decades ago: "What makes a computer a computer is our way of thinking about its potential... We are the only measure of the existence of computers." Many bits from that paper are peppered throughout this essay, hopefully among some of my own, too.

 

 

DOMINO BRAIN⁶

 

We conceive of computers as electronic devices, but their logical gates can be implemented in cogwheels or even lego blocks. According to something called the Church-Turing thesis, as long as a computer fulfills certain conditions, it is considered functionally equivalent to other computers.

 

So you can run your alarm clock program on a desktop computer, mobile phone or even a thingamajic assembled out of used scratch cards, as long as you have the patience to figure out how to implement it.

 

The domino block machine we just built is really too simplistic to run much else than an alarm clock or metronome, so if we want to challenge the titanic supercomputers and run any program (within spatial and temporal limits) we need SCALE.

 

If we build a domino block computer that stretches across the galaxies, housed in an even bigger building (that has futuristic gravity technology that holds the dominoes on the floor), we can use it to run an artifical intelligence program. Let's name this computer DomPal (after the finnish words for domino and block, domino and palikka).

 

Let us now imagine that brain scanning technology has reached its apex and we can model brain activity down to every last neuron in real-time. Let us assume that neuronal connections are all that's required for human consciousness (a common and dubious assumption in AI circles - the brain is much more than just neuronal connections, and even though a brain is required for humanlike consciousness, it is not obvious that is sufficient).

 

I hop into the brain scanning device. I read a book, talk to the nurse and listen to music as the machine's rays penetrate my brain. After a few hours my neurons and their connections have been modelled and turned into a string of bits.

 

According to the Faithful, information has its own existence, so if the information structure of my brain (which is assumed to be conscious) is replicated in some other form, that other form is also conscious. Thus we can create my consciousness with dominoes.

 

Google`s director of engineering Ray Kurzweil and the Oxford philosopher Nick Bostrom are probably the biggest cheerleaders of this mind uploading manoeuvre. The CEO of OpenAI, Sam Altman, believes that his mind will be uploaded to the cloud in his lifetime.⁷

 

 

GALAXY BRAIN

 

Let's print out the sequence of bits representing my brain onto an unfathomably long piece of paper. Is this manuscript conscious? One of the Faithful might demand that those bits be run on a machine. Apparently information must move before it is conscious.

 

Let's do it. We run my brain on DomPal (or we run a brain simulator program with the model of my brain as its input). We choose the state of me listening to music. DomPal reads the ones and zeros and prods the dominoes accordingly. Is DomPal now conscious? If so, where does its consciousness lie? Is the experience of hearing harmony in the blocks themselves? Has it always been there, even when I played with them in my childhood?

 

One of the Faithful might reply⁸ that is in the totality of the particles making up DomPal. Perhaps all particles are slightly conscious, and in some configurations this consciousness accumulates and takes on more complex forms.

 

We pump the air out of the building that hosts DomPal. The space is now a vacuum. Every domino contacts another only in the act of falling, either touching the one in front of it or being touched by the one behind (remember, the dominoes bounce right back up). Every block "knows" only its predecessor and follower, and neither at the same time. There is no physical chain of contact longer than this.

 

The computer functions exactly as before, but where is its consciousness? To emphasize this point we make the dominoes ever taller and thereby we can lenghten the distance between them even more (they could be miles high!).⁹

 

We can make the situation even wackier. Let's drill the domino block's hinges tightly to the floor. We turn the dial of our gravity generator down (good thing our dominoes can't float off). We set the prodding mechanism of DomPal to be very gentle - a soft and gentle prod. The dominoes now fall veeeeeery sloooooowly.

 

Mankind could destroy itself in a nuclear war and climb back out of the ashes into a flourishing civilization before DomPal could finish running any program. Does my block-brain experience time as "quickly" as before?

 

If I didn't know any better, I could assume that DomPal had stopped entirely. Is there a threshold after which the blocks cease to be part of a computer and are just almost-immobile blocks? In other words, what's the difference between an agonizingly slow computer and one that has shut down? How would you tell the difference?

 

If a computer can be conscious, then there is no reason why a computer that takes a billion years to run isn't as real as a computer that takes a second (as long as the computer holds itself together, all computers break down eventually). If DomPal were to be razed, it would be equivalent to murdering me.

 

The Faithful may consider all of this too absurd to be worth thinking about, but then new problems dig themselves up out of some crevice, like termites after the rain. Let's ponder what something smaller than a termite, a microscopic blob of bacteria, might observe. The functioning of my brain might be just as absurd to it as DomPal is to us.

 

If we do not entertain the existence of absurd computers, we inevitably put the computeriness of our brains into question. Absurdity is in the eye of the beholder. There may already be countless computers in nature so absurd that we cannot even imagine them.

 

When we cannot objectively detect what isn't a computer, the definitional power of the term "computer" is weakened. If everything is a computer, then everything can potentially be a mind, if the mind is a computer.

 

What we sorely need is an objective computer-detector, but since such a thing doesn't exist, we must rely on human interpretation. This is why making the computer the cornerstone of one's reality is a bad idea. It leads to philosophical programming errors, also known as bugs.

 

 

A BIT IS A BIT IS A BIT

 

The physical world may seem somewhat cramped to the Faithful and may require feats of pretzel-like intellectual flexing to sustain. If they accept the theoretical possiblity of DomPal, they must now explain what consciousness-producing physical processes my brain and the rows of dominoes share.

 

To remedy the situation they may tack on new mechanisms and conditions for consciousness ("only computers that follow rules X/Y/Z can be conscious"), but at the same time they restrict their space even more and destroy the explanatory power of the computer. There may be so many conditions that only the brain can fulfill them.

 

Some of the Faithful are ready to jump into the deep end. They see information processing as the fundamental ground of the universe. A row of dominoes is as much a bit as a neuron in the hands of the right computer. There is no difference between DomPal and my brain, and my mind is to be found in both.

 

But where else is my mind squirelled away? The structure of my brain is converted into bits according to certain rules. In the same way, we can find rules that make the wheat of the fields, the droplets of water in a storm or the flight paths of birds into a similar information-doohickeys. We just have to interpret them the right way. Does this make them conscious?

 

The Faithful may retreat a few steps here and defend themselves by pointing out DomPal's causality. The domino blocks actually effect each other physically, which makes them real and conscious (this is a sort-of middle ground between physical and information-based worldviews).

 

For example, a rain cloud could not implement my mind. Even if its droplets were related to each other in some absract mathematical way that mimics my neuronal connections, they would not effect each other in the same way. It wouldn't be really real.

 

This is not as watertight as it may first appear. Causal relations have to be isolated in some way. The dominoes effect each other, but the dominoes did not appear out of nowhere. Are the parties that set up the dominoes and their ancestors part of the machine as well? How does consciousness "know" to restrict itself right at the edges of our domino contraption?

 

Phrased in another way: DomPal is a computer, but it also contains computers (like my brain contains little sub-brains or regions) and is part of bigger computers (like we are parts of our societies). Do we now have an endless parade of nested and overlapping consciousnesses? Do I and my phone and my room share numerous fields of consiousness?

 

A bit doesn't know what totality of bits it belongs to or what computer is processing it. A bit can be a part of an endless number of information structures that an endless number of computers are processing. And since a pile of blocks or a neuron or a transistor can be a bit, even a whole computer can be a bit. 

 

 

THE PROGRAM OF THE SOUL

 

A rainstorm might not implement my mind, but a meteor storm might be a better candidate. We can imagine how in the vacuum of space meteors would exert a gravitational pull on each other (this example is straight from the Lanier paper).

 

For a moment of time this meteor storm could be an exact copy of DomPal. The asteroid swarm simulates DomPal, which simulates my brain, which implements my mind (or maybe the order is reversed, maybe brains are just whatcamacallits that simulate different asteroid swarms).

 

A moment should be enough. Computers come and go just like human beings, who are assembled out of physical particles that quickly dissipate.

 

There are computers implementing me everywhere, potentially at least, as long as we look hard enough. Even if my physical body perishes, in the future some other slice of the universe will implement my mind, if only momentarily. And who knows, maybe some slice already did, long before I was born.

 

The ghost of my consciousness is projected across the universe. It and all its variations and combinations live outside of time and space, looking for suitable computers to incarnate into. From this follows that all possible programs live "there" as well. Even those that never find a computer to call home.

 

The Faithful can always draw up new rules, but they tend to leave an artificial aftertaste, because they are not derived from first principles. Their only purpose is to prevent the collapse of a computer-centered worldview. And in my judgement, they cannot escape the fact that objectively detecting a computer requires a divine perspective, from which we could witness all causal relations and know the only right interpretation of them.¹⁰

 

 

A BAD DEAL

 

I am not arguing that human beings should be classified as supernatural or that we have some objective, special position in the universe. I just wish we had the humility to recognize our own ignorance about the nature of the mind.

 

A computer is an idea born out of the human mind¹¹, and no idea perfectly represents reality. At least not when bound by human limitations. When we shackle our self-image to a concept we thus restrict ourselves.

To make a computer conscious we have to make persons into computers. This is what you call a bad deal, especially since computers don't exist.

 

The reader may think I'm being facetious. After all, I could say that brains don't exist, that they're just a concept invented by human beings. But brains have a tighter definition and a physical structure that computers don't. Everything can be a part of a computer, but not everything can be part of a brain.

 

Let's forget brains for a second and push the counter-argument further. If I say that human beings are just a concept invented by human beings, the weird qualities of the mind begin to bubble up again. A human being can recognize a human being and a computer, but a computer cannot even start to play this game, because it cannot recognize itself or others as real.¹²

 

An oversimplification might be practical here: I find it more plausible that a computer is a concept invented by humans, than humans being a concept invented by computers. Everyone can pick their own side, there is no certainty about this.

 

Computer science is a wonderful and extremely useful lens through which to look at the world. But it is not apparent that it is the master key to the understanding of reality. A computer's functioning can be explained without computer science.¹³ Brains, too, chug along the paths laid out by physical laws. However, my belief in atoms is stronger than my belief in computers, as we can detect atoms with scientific instruments. Still, physics has no explanation for consciousness either.

 

But still, we know we are conscious¹⁴, we can be certain of nothing else. It would be smart to keep this astounding mystery at the center of our worldview. That is the most natural place for it to sit, unless you want to demean the only thing that you can directly experience: yourself.

 

The Faithful are all too eager to diminish the importance of their consciousness. This self-flagellation is seen as a virtue. Since no one knows what consciousness is, it must be an emergent byproduct of computation. Any other kind of relationship to experience is hopeless sentimentality and denial of facts, cowardice in the face of an inevitable truth that only the Faithful have courage to confront.

 

By squashing the mystery of consciousness they only find it again in other forms. You can't escape it.

 

 

A POWER DRILL APPROACH

 

The need to create an objective entitity, free of human context, reflects the universal religious impulse to escape human fate. Artificial intelligence promises to free us from the difficult responsibility of understanding and making sense of life.

 

One can sense a disgust towards the body and temporality in Silicon Valley's worldview. The mind is seen as separable from its physical boundaries, and its essence can be transferred from one physical implementation to another.

 

The body on the other hand is a prison that humiliates the pure mind by shackling it to physical limitations. The Faithful do not want to just push these limitations, they want to transcend them entirely.

 

The medieval concept of humanity is recast in the digital age, as man basks in his misery and misfortune in the face of AI. Geoffrey Hinton, dubbed "the Godfather of AI", who has recently been on a PR tour¹⁵, compared humans to caterpillars that extract "nuggets of understanding". The butterfly? OpenAI's GPT4.¹⁶

 

The hardcore Faithful speak of a Singularity, the End Times that Silicon Valley is destined to bring about. At that point AI becomes so powerful one might as well speak of gods. 

 

The AI's will then solve all of humanity's problems and make the chosen immortal by uploading their brains into bitspace (as we call cyberspace in Finland).¹⁷ The heretics, who dared mock the computer, will finally face its wrath.

 

I believe this religious impulse partly explains the dominance of the computer metaphor in our culture.

After all, it does place a group of humans in a very special position if they are building the God-to-be.

 

I'd just like a power drill. A person could do a lot with one of those.