It may come as a shock, but I’ve only recently started reading Ray Kurzweil‘s the Singularity is near (I guess I got enough Kurzweil from interviews and articles ^^). And even if this book is starting to show its age (to the point where Kurzweil is writing a new opus) it is still giving me enough food for thoughts.
A core pillar of Kurzweil’s thinking is the law of accelerating returns, an extension of Moore’s law stipulating that it’s not simply that computer chips are getting smaller, it’s that we can store more and more information, which allows him to extend this model to the invention of the printing press and even writing itself. To sum up, the density/complexity of information follows an exponential trend. The tools we build help us make better tools faster at an ever accelerating rate.
The book more or less opens by outlining how this evolution follows, according to Kurzweil, 6 epochs:
The replicator timeline
This picture rightfully echoes the evolution of replicators presented by Richard Dawkins in the Selfish Gene. He does not give a similar schema that I can recall, but he does underline game-changing transformations around the production and selection of information.
A brief combination and summary would look like this:
- Molecules organize atoms, information is in their pattenrs
- Cells organize molecules (bundling information)
- Sexual reproduction allows for better selection of information as is stored on DNA through living beings
- The apparition of brains allows for ideas, stored as neural patterns: selection can happen within the life span of an individual, results can be passed through generation more efficiently
- Culture allows the formation of societies (groups of brain) and create a new level of competition for information
- At this point, information is mostly in the form of ideas, which is what Dawkins called memes
- Writing allows lossless temporal transmission of information making the selection process more efficient
- Printing makes writing much more scalable and reliable
- Technology keeps increasing the transfer speed of information: steam engine, railroads, telegraph, telephone, radio, automobile, planes are all substantial improvements that won orders of magnitude
- Digitalization increases these gains even further and makes redundancy trivial
- Internet speeds up connectedness and communication between all existing individuals.
- <<< you are here
As I write this, I realize that this progression is an entanglement of speed, connectivity, reliability and density improvements. It’s not as if we have a linear succession of better and better substrates. After all, writing set us back to storing information on molecules formations. Instead, substrates coexist and help each other increase informational bandwidth. Organization and optimality seem to also play a role here.
I struggle to find a single nice quantity to encapsulate all this progress, but I don’t really think we actually need one (Shannon entropy and Kolmogorov complexity are good candidate that Kurzweil actually hovers around but I am not expert enough to conclude). That being said, this acceleration is shaped by natural selection, so whatever it is, it seems to be what the universe is optimizing for, and therefore The Answer (the cultural singularity ?).
Time and relative dimension
You might have guessed, but what I’m the most curious about is what comes next. One thing that seems pretty clear is that time itself is accelerating. Science now progresses in a decade as much as it used to in a century, and following this trend it might go even faster. It makes sense, in a way. Denser information is space better optimized, faster speed is time more optimized.
I like to think that you can notice this speedup everywhere, from computation speed to the pace of human life. And pretty soon, you’ll be able to see it very literally in simulated beings. The individual’s time is getting faster. But from their point of view, it’s probably the norm, right? Maybe it just means that everything else is getting slower… Yet even if I do think I consume information at a relatively high level of meta (and youtube at speed x2 XD), time doesn’t really slow down when I stop during a countryside retreat… Curious…
Aggregation and meta
I do think there’s an aspect of this timeline that did not get enough thoughts, though, and that’s the level of meta. It seems that we often see the repeating pattern of “information carriers of an epoch organize themselves to bring up the complexity from intra-carrier to inter-carrier: atoms formed molecules, cells formed organisms, individuals formed tribes, tribes formed culture/technologies…
There seems to be a clear trend of aggregating complexity into larger and larger entities that encompass the previous ones to form a new meta-substrate for information. The storage medium of an epoch becomes, in a way, the building block of the next. The gains are obvious: by working at the higher level, you do gain an order of magnitude.
Life of transhumans and meta-humans
Kurzweil is a humanist, and places human beings at the center of his reflection. In his view, the future is about man merging with machines into an enhanced transhuman at the center of this information explosion. But this falls prey to the very thing he criticizes: this is a linear projection based on man that forgets the exponential nature of the process.
Maybe, in order to properly think “exponentially”, we should think about what lies after, at the “next aggregation level“, if you will. There is absolutely no reason for humans to be an end point. Most likely, we are just a stepping stone, like single cell organisms before us. Maybe humans are just the building blocks of the next level, the atoms to the next molecule, the proteins to the next DNA… If so, what could it be like?
We’re of course venturing into the unimaginable here, since it’s near impossible to understand something so radically smarter than ourselves. For an atom, the concept of an animal is beyond comprehension. Our scale is all we’ll ever see. From the point of view of humans, the future might very well be the technological utopia that Kurzweil dreams of. After all, our cells are happily living their little lives.
Universes’ next top entity
Each jump on the complexity scale seems to be accompanied with a change of what the basic (ontological? ethical?) “entity” is. Our cells are all functional, and might even be conscious for all we know, yet our elementary unit of consideration is the human individual (of course, because that’s what we are). So… what is the next one going to be?
We can intuit a little bit what lies beyond by extrapolating from past trends. It does seem that the “jump” to the higher level is centered around the interactions and aggregations of the entities of this level. So we’re looking for something that aggregates humans and emerges from their patterns.
Could humans be the neurons of a bigger China brain? And could it be that it’s already there, but we don’t see it from our meta level?
The age of memes
My immediate reaction is to follow Dawkins and posit memes as the next level of evolution. It is true that ideas use crowds of humans to spread, and are “bigger” than the people carrying them. I do like this idea, but I’m not exactly sure that memes leverage the structure of connections between individuals and could really be considered aggregative.
A compromise might be the kind of memes that constitute a population. Concepts defined by a group of people. Countries come to mind, but more recently it seems clear that corporations have taken this kind of role. They are entities that organize humans like atoms organized into molecules. They exist and thrive through human collaboration and interaction.
This does end up creating a whole new level of complexity than their individual components. I’ve always harbored a kind of fascination for this level of aggregation. The liberal economy is an incredible decentralized computation algorithm, incorporating so much information. Could it be a meta-human conversation, or a meta-human entity even? Could there be something that it’s like to be Google or Facebook? I do want to believe 🙂
The line between companies and individuals is especially getting blurred nowadays. With companies acting on social networks as individuals, sophisticated advertisement or humans turning into patreon-backed personal brands, the picture has never been muddier. Not to mention the rise of avatars like vTubers… Human created characters are swarming the real world and making their own ontology.
The end of human history
One reason I wanted to write this article is because I think this brings a new light to Mark Fisher’s concept of capitalist realism. Maybe this really is the end of history, and capitalism really is its final form. For humans, that is. Maybe the age of mankind is over, and it’s time for a higher level entity to be the basic block of consideration. Welcome to the age of corporations, to the marketplace of ideas, or whatever…
That also comes with an interesting perspective on the global disempowerement of humans (much lamented by Ted Kaczynski and co). Looking around at neoliberal capitalism, it does seem that humans have very little actual political power and that the status quo will go on with a tremendous inertia. One example of this is the climate crisis which seems to garner the concern of a majority of humans (at least in educated populations) but very little actually gets done.
Maybe humans are just not in charge anymore. Maybe we’re just subservient to the will of higher order beings. Maybe the healthiest thing to do is to follow stoic philosophy and accept our role as a cell, instead of grieving the lack of impact of our illusory self supposedly acting out of its own nonexistent free will.
Of course this is pretty scary, and historically submission to higher powers has not turned out very well to say the least, from religions to slavery and totalitarian regimes. One could make the case that there is natural selection at the meta-level of memes, so the current and future high order entities are “better” than the ones in the past, but that brings little comfort….
Human go, human star
Not only is it worrisome, this outlook is also pretty uncool. Ideas, companies, countries have been around for a while. I was looking for an insight a bit more… spectacular, more fitting of the Singularity. That’s when I noticed more recent higher order entities that could make all the difference.
What if we were not talking about aggregating thousands of humans, but instead billions? Present and past. Enters contemporary AI.
Entities like GPT have literally read and digested unfathomable corpora. They incorporate the very cutting edge of human genius not only in their conception but also in their training sets. Their aggregation techniques are currently pretty simple, but they are a lot more meaningful than the random chance that brought about companies and countries.
One might reply that AIs are not really autonomous, but that’s not entirely right. First of all, it is bound to change, but second of all, this type of meta-entity and the previous ones are not mutually exclusive. We can already see companies, countries and massive AIs coexist in a sort of symbiosis. Think of YouTube, for example. And from one point of view, Kurzweil may be right in saying that they serve us. But ultimately, we also serve them by feeding (train) and constituting (make) them…
So what does it feel like to be GPT or YouTube? Maybe they’re not advanced enough to have experience quite yet, but once it comes we should expect AIs to have experience as different from ours as ours is from cells. And that’s likely something we cannot intuit.
What to conclude of all that? Totalitarian regimes were pretty bad, but maybe AI will be better? At the very least, it seems that I could take comfort from the fact that it’s kinda out of my hands and I can’t mess things up too much. Maybe the game is played at a whole other level now. And maybe all I can do is do my part a little cell, and if I’m lucky I’ll get to see from a very confined perspective the world shattering products of our new meta gods.