This post was written by @senyakk and is cross-posted from our Substack. Kindly read the description of this sequence to understand the context in which this was written.
An AISIG colleague of mine, @ilijalichkovski, published a commentary critique of Mechanize, an AI startup manifesto, which appears to be a defense of technological acceleration due to its deterministic stance. I refer to the manifesto’s proponents as the mechanists and to my colleague, respectively, as the author.
The Mechanists advance two theses: 1m) the tech tree is discovered, not forged, and 2m) we do not control our technological trajectory. The author counters that what mechanists overlook is that determinism can still be compatible with the historical contingencies – a valuable correction to fatalism – ultimately defending the view that 1a) the course of history is conditioned by the power law, and that 2a) the outcomes of the technological tree in fact diverge and not converge.
I find the author’s arguments prone to supporting a position leaning towards a voluntaristic interpretation of history. From the premise of a power law, one could infer that the society is governed by the same power inequality, with different people’s acts having different weights. Given the indeterminacy of the future, voluntarism attributes the course of history to the will and wisdom of individuals in power, rather than to impersonal structural forces. In this essay, I will argue that automation is indeed inevitable in the history of humanity, but what curtails human agency is not this inevitability but the socio-economic inequality that conditions the application of automation technologies.
To understand the course of history, I’ll define history as the social process that consists of events and phenomena, connected by causal relations.
Since the time of the Enlightenment, the dominant idea in historical science was of a progressive march of human knowledge, having its own objective laws of development, a view appearing in the works of Turgot. The empirical basis by that time made it clear that society was not stuck in a loop, but instead was advancing through periods; Adam Ferguson famously categorised human history epochs into “savagery”, “barbarism”, and “civilisation”.
With this in mind, the definition can be expanded as “the objective social process”. “Objective” here is meant in the sense of independence of this process’s existence from any individual in particular, albeit, of course, this process can only exist through the totality of all individuals.
The author draws an analogy:
Just as AI startup founders today make the case that total automation of the economy is inevitable, a peasant from the Dark Ages would be making the case that feudalism is the inevitable order of human affairs.
But the crucial difference is – a peasant is certain that feudalism will persist and the current order will prevail, while mechanists claim that things will keep inevitably changing in the direction of advancement.
The next logical question is about the source of those historical laws. Political Economy thinkers in the works of Smith, Ricardo, and Marx have answered that production has the leading role in the historical process. Humanity needs to produce material, and later in its history, intellectual and cultural goods to survive and reproduce itself. Crucially, these laws of development are independent of the individuals in power. Thus, I have to agree with the first thesis of mechanists. The tech tree is discovered and not forged due to production being anchored in material necessities. We are forced to advance technologies within given natural laws, and although we could master them and turn them to our favour, we can’t change the reality of these laws. This allows to conclude that the historical process is being paved independently of individuals, but is inherent in the dynamics of humanity itself.
Having established that production and material necessity ground the historical process, the next question is whether the technological evolution that follows it converges on universal outcomes or branches into divergent paths.
While it is true that, as the author claims, the particular conditions of the spread of Catholicism in Europe were precursors to the scientific revolution, it doesn’t make a case for arguing that similar advances were not inevitable through other means. The scientific revolution was a milestone of human cognition, enabling it with methods to create a more accurate, correct picture of the objective world, and this step was necessary for humanity’s growth and reproduction. Every technology is made to serve a particular purpose. Each purpose, being rooted in the objective world and human needs in this world, can only be rightly and most efficiently satisfied in the process of cognition of the world, obtaining knowledge of reality. Different technologies with the same purpose converge in the sense that humanity develops a technology that suits the purpose better than before.
The movable type printing example and its “rediscovery” centuries later, in my opinion, only speaks in favour of the idea that this was a necessary technology for humankind, since it conditioned and skyrocketed enormous advances in the years that followed. The potential of that technology, its take-off, however, was dependent on the historical contingencies, such as the use of the writing system, the material, and the context. Another notable example is the first navigable submarine built in 1620 by Cornelius Drebbel, which failed to attract enthusiasm, and those underwater vehicles were not used in combat until almost two centuries later. Drebbel’s submarine was simply too advanced for its era and lacked the necessary scientific or industrial preconditions to sustain it, like enabling propulsion and prolonged life support.
The length of the period of technological divergence is contingent, but the convergent technology offers benefits superior to its predecessors. Through contingencies, the progress gushes out of the historical process. Mesopotamian clay tablets and East Asian calligraphy both served the purpose of preserving the language, and historically humanity did converge on a single paradigm –the movable type printing. In its turn, those “local optimas” have converged on a digital typography, enabling the transmission of the entirety of the Chinese script with ease, something that its precursor lacked.
Connecting that with the idea of historical laws, this, importantly, also retains the humanist outlook, namely by acknowledging the capacity of the human mind to develop oneself and transform the world around it. As we bend to our will more domains of this world, the freer we become from the circumstances of our lives that were imposed on us by nature, and the more possibilities for self-development open in front of us.
All of this leaves a looming elephant in the room. Do we really have no say in the face of this fatalism? If technological convergence resulted in everyone’s prosperity and equal access to benefits, there would be no reason to say anything. However, the second mechanists’ thesis, that we do not control our technological trajectory, is where the subtlety arises.
While macro trajectory might not be controlled, what can be controlled is the method, scope, and extent of applications of these technologies, broadly the social consequences of the technological trajectory. Humanity is driven by production and progress in knowledge that it causes, but the choice of how to steer these achievements in any historical moment is contingent. Those are the concrete choices about where, how fast, and for whose benefit a technology is deployed. I will build on the author’s claim that those are the specifics that can lead to drastically different outcomes*.*
The mechanists’ premise hinges on the assumption that striving for economic advantage, as the primary goal of society, is beyond our control. The extent of humanity’s capabilities to influence society itself is conditioned by the mode of production. If production is the driving factor, the mode of production is the constraint. How it is being produced, who benefits from it, and for which purpose it is made are the fundamental questions that should be asked. It is then evident that what rips the agency from the majority of the people is inequality in their ownership and control of these technologies, and by extension, the inequality of power. The author is correct in pointing out that a handful of people in a handful of circumstances do have an outsized influence on humanity.
The introduction of AI automation technologies offers tremendous benefits for businesses, including reducing expenses by automating tasks, followed by firing workers and forestalling workers’ discontent and strikes by laying them off. Unionized workers are rightfully protesting against employers denying them their only means of sustenance. The automation isn’t the culprit itself; everyone would benefit from doing less unwanted work and augmenting their labour. The crux of the matter is that in the current mode of production, automation makes workers’ labour force increasingly redundant to the market, causing even more wealth to be siphoned upwards. The higher concentration of ownership exacerbates the shift of wealth from labour to capital as automation advances.
I have outlined that humanity is driven by the objective laws that necessitate it to produce and reproduce itself. Technological advance is a march towards better knowledge of the world, serving the reproduction of humanity in a general sense. Autonomous agents are the continuation of the progress in the field of computing, which, like any other scientific and engineering field, serves the purpose of knowledge and making our lives better, empowered by this knowledge. Mechanists are right in that the transformative technologies will be developed anyway. Humanity will inevitably come to autonomy, unless it is destroyed or rolled back into the Stone Age in the fallout of a world cataclysm.
Automation is inevitable, but the distribution of its consequences is not. Any technology primarily serves the interests of the owners and investors. Society is fundamentally split at its level between people, or rather classes, with different economic incentives. What remains as our “only real choice”, to quote the mechanists, is not to hasten technological progress, but to consider the power and class inequality relations present in society and to take steps towards humanity taking full material ownership and control over the technologies it produces. History will advance regardless, but who commands its fruits — that remains entirely within our hands.