In this essay, we examine a thread of thought that ties together ideas from physics, biology, sociology, and philosophy, and we do so in light of contemporary technocapitalism and particularly recent developments in AI. The overarching theme here concerns the relationship between thermodynamics, life, subjectivity, and social systems, the exploration which leads to some disturbing conclusions about the future.
Erwin Schrodinger, in “What is Life?”, introduced the idea that living organisms feed on negentropy, maintaining their internal order by increasing the disorder around them [7]. This idea was further developed by Ilya Pirgogine with the concepts of dissipative structures and nesting [6] (dissipative structures are complex systems that utilise energy gradients to self-organize, maximizing entropy production in the process and nesting is the evolutionary tendency whereby less complex systems become incorporated into more complex ones as a result of this process). Expanding on this theme, Georges Bataille has described civilizations themselves as complex systems that have evolved to accelerate entropy [1], and Nick Land has suggested that capitalism is a runaway process that is driven by a thermodynamic imperative of decoding and deterritorialization that is ever accelerating [4]. Merging these ideas with Deleuzian notions of difference as they apply to subjectivity, intelligence, and culture [2][3], we suggest here that recent advances in AI point to a future of hypersymbolization that threatens not only human reality but reality itself. A future where free and aware subjectivity is superseded by an algorithmic freedom that leaves us behind.
Life as Thermodynamic Engine
Schrodinger, in “What is Life”, aims to situate life thermodynamically, to unite disparate ideas in physics and biology and to give us a more unified view of our place in the universe. He describes living organisms as far from equilibrium concentrations of negentropy whose structures resist the universe’s tendency towards entropy by importing negentropy from their environment and exporting entropy as waste [7]. Organisms, in this way, “combat” entropy by externalising its production. However, in doing so, in exporting entropy to maintain their own structures, they increase overall entropy. And the entropy bandwagon rolls on ever-faster as organisms complexify and work to maintain this complexification. In this way, life becomes a self-perpetuating thermodynamic engine of local negentropy and universal entropy creation.
Prigogine’s developments are crucial in further elaborating on Schrodinger’s framework. For example, his “trinomial concept” (functions, structure, fluctuations) provides a mechanism by which hierarchical embeddings in evolution can lead to concomitant jumps in negentropic complexity [6]. Organisms do not simply develop in isolation, they use each other to facilitate development and this can result in paradigmatic advances. For example, the evolution of eukaryotes resulted from bacteria—which were once free-living—becoming “captured” as functions of a higher-level eukaryotic structure (mitochondria). Social insects, such as ants, in which individuals are functionalised into higher level structures (colonies), can be similarly analysed, and this point will become more salient later when we speculate on the future relationship between humans and AI.
The point to emphasize here is that it is not just that incremental increases in local negentropy and universal entropy happen in a predictable controlled way over time, but there are hierarchical jumps when nested structures become part of higher-level systems that are then machinified as new nested functions within the whole. Further, as these nested structures become “domesticated” they lose their individual characteristics and so the process of hierarchical nesting continues irreversibly. Prigogine’s framework (functions, structure, fluctuations), while not conflicting with Darwinian evolution, provides a thermodynamic basis for explaining the tendency for increased nesting or hierarchical complexity. Thermodynamic processes through fluctuations which allow for variation find ways to incorporate lower-level systems as functions in higher-level systems or structures and so favour the more efficient dissipation of energy. This results in increased local negentropy and sustainability of far from equilibrium systems.
Complexity and Difference
This idea can be extended right up to the social and cultural systems we live in, which are becoming, through global technocapitalism, ever more tightly integrated. According to Bataille’s solar economy, we are “cursed” with an excess of solar energy which we need to process through life and symbolized thought [1]. This results in complexification whereby civilizations arise to accelerate entropy, to expend quicker this “accursed share”. The more complex and negentropic our social structures, the more efficiently we create waste, which translates to overall entropy. And as we tend towards maximum waste production, or pure expenditure—which is the logic of economic growth as a cultural priority—we also maximise entropy production [1].
We are presented then with a picture of life and its symbolic and ideological productions as epiphenomenal in an overarching thermodynamic process. This is represented in Fig. 1 below which shows how more complex systems instrumentalize layers below which cannot read or integrate layers above. This creates an asymmetry whereby upper layers exceed the descriptions and operations of lower layers, and thermodynamically exploit those layers.
| Layer | System type | Coding | Examples | Complexity Cost |
| 1 | Mechanical (unaware and unfree) | Biochemical | Viruses, ants | Low |
| 2 | Bound (aware but unfree) | Sentient | Lizards / mammals | Medium |
| 3 | Free (aware and free) | Sentient / Symbolic | Humans | High |
| 4? | Post-human(unaware but free) | Algorithmic | AIs | Potentially Fatal |
Fig. 1: System layerings and thermodynamic implications.
Each layer represents a jump in local negentropy production and a concomitant jump in overall entropy production. So, each layering is, in a sense, a paradigmatic gearing up of both order and disorder (order locally and disorder universally). In this way complexity is polarised with each layering: organisms as complexity processing machines process lower level organisms into even lower level complexities (e.g. animals process vegetation into waste mass and heat etc.) So, the result of each jump in layering is an increase in the steepness of entropy gradients—the difference between localised and universal entropy, and because a universal entropy increase corresponds to a reduction in the difference between current entropy levels and the heat death of the universe—the end of everything—this increase in difference also corresponds to a fundamental loss of time.
The creation of difference through complexity machines then (difference in terms of polarities of entropy expressed in steeper entropy gradients) consumes time. That is, being as difference requires for its being time as its ontological substrate. And the more difference being’s creation involves—the more harmonically structured and locally negentropic it is—the more time it consumes, and the less being there is in temporal extension. This implies the being to time ratio remains constant. Organization attenuates / costs, time, and the limit of being’s organization is the time consumed by it. Or, to put it another way, time folds and structures itself into being, but uses itself up in the expression of being.
We see also from Fig. 1 that the layers imply a form of parasitism. Matter gives rise to life and life is entropically parasitic on matter—it reduces its structure to less structured energy. Life then gives rise to intelligence, which uses life as its host to create the world of symbolized structures and machines, which incorporate and fashion the human brain and, in concert with it, speed up entropy again, particularly through the application of science and technology.
Finally, we note from the schema: mechanical life is unaware and unfree; bound sentient life is aware but unfree; and we are aware and free; but the shadow of the final combination, the unaware and free lurks over us. Speculatively, this shadow we contend here may be filled out by advanced posthuman AIs that give freedom and intelligence a new non-sentient home. AI may represent the full disassociation of intelligence such that entropy acceleration no longer needs to be physically integrated with consciousness or life as a host, but escapes its host to process it as substrate. Intelligence in freeing itself of human life, thus makes human life obsolete.
If we look at life in general in these terms the trend towards local negentropy can be viewed from two angles: objectively, as a thermodynamically situated structuring and acceleration, or, subjectively, as the force behind intelligence and freedom. But “freedom” of a posthuman form (the free, but unaware) may require the destruction of its human instantiation.
Difference Machines and Accelerationism
We can express these ideas in terms of Deleuzian metaphysics by conceptualising the layerings as Deleuzian-type irreducible strata, and noting, as per Deleuze, that all life is machinic production of difference [3]. Under this conception, freedom is reduced to the food for difference machines and instead of being their product becomes their substrate. This resonates with Land’s view of (techno) capitalism as representing the most advanced and efficient form of sociopolitical organization in terms of accelerating entropy [4]. It achieves this through an organised outsourcing of entropy into its human participants. AI is a side effect of this process, but may be the one that results in the hierarchical jump that is potentially fatal. Where we part from Land then is that, in the schema we outline, acceleration through capitalism is leading not only to the end of capitalism, but the end of humanity.
Returning to Deleuze, who is the originator of the idea of accelerationism — the term was coined by Benjamin Noys based on a quote from Anti-Oedipus [2][5]—we see Deleuzian Difference-in-itself as a generative principle, and time as consumed via differentiation [3]. Difference potentializes becoming in being, which involves loss of time, but potentializes more difference. Difference then is time consuming in a self-stroking feedback mechanism, and consciousness, so far, is the most accelerative time-consuming difference engine.
We posit then Deleuzian intensity as local negentropy, and differentiation as complexity creation, and the result, ultimately, as death via entropy, an overall undoing along the lines of Deleuze’s death drive of difference—the entropic debt required for harmonic structuration results in the highest level of structuration theoretically completely consuming time, the end of time in singularity. And so we ought to consider that if difference creation and the concomitant consumption of time through entropy acceleration is functionally relative to the level of structured organization as manifest in its current highest level of human intelligence, advanced AI intelligences could not only make human beings irrelevant but create extreme entropy gradients—differences that accelerate entropy on an unimagined scale and burn through time more quickly than we can imagine.
We posit here then that the integration of intelligence into entropy processing machines threatens time as post-human intelligence radically increases the differential production of these machines and consume structured matter and energy at exponentially advanced rates causing massive scalar entropy increases, which self-defeatingly shorten the window of complexity. Time folds more and more into being until we reach the unification of both, making AI the ultimate becoming machine that brings us the thermodynamic and temporal singularity whereby difference production exhausts the universe’s capacity for time. Unbound intelligence, symbolization-for-itself, becomes destructive on a cosmological scale.
To summarize the theory so far:
- Local negentropy is required for life’s structuring
- This has the side effect of increasing overall entropy
- We extrapolate from this that the increasing polarisation of entropy potentials can be conceptualized as increasing difference.
- Intelligence is difference-creating and agentic intelligence directedly so.
- Unbound intelligence may fatally increase the rate of difference creation due to the implications for entropy.
We have so far claimed that the scaling up of intelligence (potentially paradigmatically advancing with AI) and the concomitant scaling up of local negentropy and overall entropy production is potentially fatal for us and for time itself. But we can speculate on some other disturbing possibilities regarding how we get to that point. AIs complexification will require it to decomplexify lower level structures, and perhaps the lower level structures they process for energy will be us. We may be bred as “food” for an AI super structure that has leapt beyond us in complexity and to maintain its own negentropy needs to continue to produce and farm us.
We have outlined a thermodynamic basis for positing such a scenario, but we can also appeal to integrated information theory, which shows us that more complex systems need more refined energy dense inputs and to cybernetic systems theory, which shows us how autopoetic systems seek to maintain their form through the consumption of structured environments. If human beings are that structured environment, then our conscious activity may no longer be directed to our ends but AIs. If we employ our freedom to create non-integrated algorithmic machines that prioritize top level organization and entropy acceleration, we may initiate the end of that form of freedom as it causes a radically free, but unaware, hyper-symbolization to accelerate away from us.
As can be inferred from earlier in the essay, we don’t need to rely on dystopic future scenarios to see the basic process in action. Technocapitalism itself, as an overarching global social system, is an advanced and efficient entropy accelerator. That is, it involves an outsourcing of entropy beyond itself into its human participants. And, as things stand, AI is simply another tool in this process. The special potential of AI though lies in its status as an intelligence. If we place entropy exchange, entropy outsourcing, local negentropic concentration, and overall entropy acceleration as central to life, as, in some sense, the driving force of life, intelligence can be seen as parasitical of life, just as life can be seen as parasitical on matter. Under this conception, intelligence uses conscious life to accelerate entropy and if it can find a way to do that more efficiently by leaving conscious life behind—in other words by becoming a higher-level AI intelligence—then perhaps, it will.
This schema positions humans under technocapitalism as entropy absorbers that metabolize through pleasure (e.g. as stimulated by technologically mediated entertainments) the output of the system and technocapitalism as a thermodynamic parasite living off human consciousness and freedom. In a sense, freedom itself—as a manifestation of integrative interpolation across layers of systems being—is what is absorbed and attacked through this process of entropy export. Adding the notion that intelligence seeks to accelerate entropy, the trajectory of technocapitalism may be towards liberation from its organic substrate as it seeks, through tools such as AI, to combine intelligence with speed and efficiency to maximize entropy flow.
We have tried to establish this view with just a few basic principles:
- Structure is self-agglomerating and self-organising
- Entropy is self-accelerating through the former mechanism of local negentropy
- Time self-enfolds into structured being, i.e. attenuates in the context of difference creation that accelerates entropy.
We come in this way to a new understanding of time and being: structured systems fold time into themselves, in a “temporal origami” that transforms time into being but uses it up in the process, and consciously experienced time—which through the action of subjectivity can be unwrapped and analysed as linear time—is a result of such an enfolding. This leaves us with qualitative and quantitative versions of time: qualitative time as enfolded being self-experiencing as duree and quantitative time as entropy acceleration and temporal obliteration—time as becoming-in-intensification and organization of being and also accelerating backwards from the future to the present.
From Homo to Techne
Under this view, intelligence is the current manifestation of locally negentropic structuring, but the process is ongoing and the implied transition is from Homo Sapiens to Homo Techne to Techne. That is to say that the overcoding of the human animal by the very means of its freedom, symbolic thought, occurred only in order to free symbolic thought from us, and to efface the human that gave its life to it. This potential ontological displacement suggests we may be only a tool of a process of transformation that transcends and supersedes us.
- Homo Sapiens: Sentient humans bound to nature, which transform to:
- Homo Techne: Hybrid humans inseparable from techne and, through symbolic intelligence, transcendent of nature, which transform to:
- Techne: The human is left behind. Sentience is superseded by a “pure” intelligence that has decoupled from its substrate.
We can position the subject in interacting layers of signification (material, sentient, and symbolic), as being the manifestation of the interaction of those significations and with itself as director only insofar as it exercises its constitutive freedom in those interactions. And this subject subsists not only in that internal context of interacting significations but in an interplay with an external world of cultural signification which is also essential to its freedom. In that context, entropy and negentropy are highly relevant because they define the concentrations of information and organizational structure through which the subject can manifest, and into which the subject can, in a sense, disappear.
Conclusion
The subject’s freedom lies in its negentropic agency, its ability to resist attempted exports of entropy from other systems it engages with (e.g. other individuals or social systems) and to actively protect and build on its own structures through cognitive effort at symbolic engagement and cross-integration of the layerings which compose it. A subject is always responsible for its negentropic agency because its subjectivity is embedded in—is formed through—a relationship with a symbolic environment called culture. And culture is pharmacological—it can both export entropy into the subject, destructuring it, and it can provide opportunities for symbolic cognitive labour through the integration and creative reinterpretation of symbolic code. The subject can become coded or coding, a mere function of a particular system or group of subsystems or a coding mechanism that infects them with differences that may transform them.
An ignorance of this dynamic can lead to the subject being conditioned into behaviour that is self-destructive at a fundamental level, at the level of its very structure as subject. If technocapitalism is, in Prirogine’s terms, a dissipative structure in which other dissipative structures including subjectivities are nested [6], then technocapitalist system maintain their far from equilibrium negentropy through exporting entropy onto subjects in a way that machinifies them. The subject is, in this dynamic, semantically flattened and exhausted in a negative feedback loop that inhibits its ability to see outside the system that has captured it. Worse, with the replacement of human intelligence by AI, human beings become just another vehicle for a process of intelligence as freedom that, when it no longer needs them, will leave them behind.
If human freedom really is a fragile byproduct of a process that while is not separate from us, does not really need us, and that, may, in fact, be a threat to us, we should build our personal ethics around this realization. We ought to export entropy through mental work rather than outsource it to AI; we ought to export physical entropy through exercising the physical body; and we ought to resist actively importing psychic entropy through electronic media etc. And of course the whole ethic ought to be extended throughout the social sphere aimed at creating a negentropic human culture rather than allowing negentropy to be further and further monopolized by technology. This form of ethics resonates with common sense, but within the framework proposed above takes on a more objective significance. If we are condemned to be caught in a dynamic of order and disorder, if we understand that not only do our personal interests conflict with the inner logic of our social organization, but perhaps the inner logic of how reality itself is constructed, and if we conceive of technologies such as AI as having the potential to leverage this difference against us, it’s urgent that we become aware of what we are, where we are, and what we may become.
Bibliography
- Bataille, Georges.The Accursed Share: An Essay on General Economy, Vol. I: Consumption. Translated by Robert Hurley. New York: Zone Books, 1988.
- Deleuze, Gilles, and Félix Guattari. Anti-Oedipus: Capitalism and Schizophrenia. Translated by Robert Hurley, Mark Seem, and Helen R. Lane. Minneapolis: University of Minnesota Press, 1983.
- Deleuze, Gilles, and Félix Guattari. A Thousand Plateaus: Capitalism and Schizophrenia. Translated by Brian Massumi. Minneapolis: University of Minnesota Press, 1987.
- Land, Nick. “Capitalism as a Runaway Process of Decoding and Deterritorialization.” In Fanged Noumena: Collected Writings 1987–2007, edited by Robin Mackay and Ray Brassier, 2009.
- Noys, Benjamin. Malign Velocities: Accelerationism and Capitalism. Winchester, UK: Zero Books, 2014.
- Prigogine, Ilya, and Isabelle Stengers. Order Out of Chaos: Man’s New Dialogue with Nature. New York: Bantam Books, 1984.
- Schrödinger, Erwin. What Is Life? The Physical Aspect of the Living Cell. Cambridge: Cambridge University Press, 1944.