By Javier Surasky
Este blog estará disponible en español el viernes por la
mañana.
When we speak about artificial intelligence (AI), we often take for granted assumptions we rarely question. Among them is the—intentional or not—assumption of an agent ontology: systems that “learn,” “decide,” or “act,” endowed with an “ethics” and “biases,” to name just a few examples belonging to the cognitive and moral domain. This language has allowed our attention to shift away from the material conditions of power toward the attraction generated by the idea of possible functional autonomy.
Yet the
problem of AI is not only epistemological or ethical—at least not in the
classical sense—but fundamentally political and structural. Hannah Arendt
(2003:16) warns that if technical knowledge and thinking become separated, we
risk turning into “impotent slaves, not so much of our machines as of our
know-how.”
There is a
profound problem of power in the Aristotelian sense. Aristotle defines dýnamis
(potentiality) not as an entity, but as a principle of relational change
(Aristotle, 1994:241), expressed as the principle of motion or change that
occurs in another—or in the same thing insofar as it becomes other. From this
perspective, power does not exist as a solid object, but as a capacity or
possibility that requires a subject (matter) and a direction (act).
In previous posts, we have examined the geopolitical disputes surrounding AI, and it has become clear that the real problem today is not what AI can do, but who controls it. We must now add a third element: who is the Aristotelian subject, and in what direction? Or, more directly, on what does the strength of AI—and its control—depend?
Methodology and analytical scope
In this
work, we adopt a hermeneutic–comparative approach in which myths are not
analysed as historical explanations of contemporary technological phenomena,
nor as symbolic anticipations of them. Rather, they are used as deep and dense
conceptual structures for thinking about those phenomena. In our approach, the
relationship established between myth and AI is analogical in the most strictly
structural sense possible, and it is valid insofar as it is employed as a tool
to problematise issues such as agency, world-making, and responsibility,
without presupposing any form of historical continuity or literal equivalence
between very different contexts.
Myth is
treated as a device of thought that serves to denaturalise assumptions which,
in technical discourse on AI, are presented as self-evident or neutral.
Following Hans Blumenberg, we approach myth as a response to what he calls the
“absolutism of reality,” referring to situations in which human beings perceive
that they have lost control over the conditions of their existence. As
Blumenberg puts it: “the ‘absolutism of reality’ designates a situation in
which human beings come close to not having control over the conditions of
their existence and believe that they simply lack such control” (Blumenberg,
1985: 3). Seen in this way, myth does not explain the contemporary world, but
it helps to make it inhabitable, symbolically delimiting what might otherwise
be experienced as a diffuse and uncontrollable threat.
The selection of the myths used here is based on their heuristic capacity to illuminate conceptual problems currently under debate in the field of AI, while explicitly recognising their limits. This approach does not seek to serve as a basis for legal norms, to design public policies, or anything of the sort. Its aim is solely to open a space for critical dialogue with other specialised approaches, by bringing to the fore dimensions of the problems they confront that should not be reduced to calculations of efficiency, accuracy, or performance.
Returning to the Classics to Find New Answers
“A classic
is a book that never finishes saying what it has to say,” wrote Italo Calvino
(2015:15), and Lévi-Strauss (1995:232) observed that “the intrinsic value
attributed to myth derives from the fact that these events, which are supposed
to have occurred at a moment in time, also constitute a permanent structure.”
Following
both authors, we will “return” to three great classical myths to seek an answer
to our initial question by establishing a framework capable of thinking about
the power of this technology as conditional, relational, and materially
grounded.
In this
effort, the myths of Antaeus, Prometheus, and Icarus appear as continuous
spaces within a single unfolding of technological power: Prometheus describes
the expansion of technical potential; Icarus speaks to technological failure
resulting from a rupture between technology and the capacity for control; and
Antaeus connects and organizes both by introducing the notion of dependence and
conditionality.
The Myth of Antaeus: Relational Strength and Conditional Power
In
Apollodorus’ account (1985:116), Antaeus is an invincible giant, not only
because of his strength but also because the dynamics of combat always end up
working in his favor: whenever he falls to the ground, he grows stronger. For
other combatants, that would lead to exhaustion, but for him, it means
regeneration.
The son of
Poseidon (god of the sea) and Gaia (the Earth), Antaeus derives his strength
from contact with the ground on which he stands. That is, he does not possess
strength as an internal attribute; it is not part of his essence. Instead, it
emerges relationally and exists only insofar as a specific condition is met:
physical contact with the ground. Antaeus does not “have” power; he can
exercise it only while touching the Earth.
The myth
tells us that Antaeus was ultimately defeated by Hercules, who, by inference,
identifies the source of his strength and applies an appropriate strategy: he
lifts him up and keeps him suspended in the air until he suffocates. Once the
circuit is interrupted, Antaeus loses his capacity to recover his strength and
becomes vulnerable. Power did not reside in the agent (Antaeus) but in the
material conditions that sustained him. Everyone who tried to defeat him “on
his own terrain” failed, because defeating him required intervening in the
factual basis of his capacity for agency—the source of his power.
Like
Antaeus, every effective form of power depends on a material environment that
gives it its character and serves as its ecosystem of reproduction. Hannah
Arendt formulates this in terms of world and action: there is no action without
a previously constructed artificial world that sustains it. Power is not
something in itself, but something that exists in relation to specific
conditions.
Artificial
intelligence represents an eminently Antaean form of power. It depends on
infrastructures such as data centers, energy networks, cooling systems, supply
chains, hardware, and massive access to data—a network that Benjamin Bratton
defines as an “accidental megastructure that is both a computational
infrastructure and a new architecture of governance” (Bratton, 2021:14), in
which the rhetoric of the “cloud” conceals the fact that AI ultimately draws
its power from the ground.
Benjamin
Bratton offers us a contemporary vocabulary to express this relationship
through his theory of the stack (or “layer”), which conceives the world
directly by integrating various layers of infrastructure and sovereignty,
anchoring the cloud to territory and digital sovereignty (Bratton, 2021:43).
Perhaps,
beyond the myth, Antaeus did not die but was transformed into an entity that
changes its form. After all, even within the mythical world, Antaeus drew his
strength from his mother, but on his father’s side, he had a (somewhat
disputed) kinship with Proteus, a marine divinity whose defining trait is his
capacity for metamorphosis—thus representing concealment and mutability.
Prometheus, Fire, and Technique
In the myth
narrated by Hesiod, Prometheus gives fire to humankind—fire understood as technē,
the capacity to produce, transform, and dominate natural processes—but not phronesis,
understood as “a true and practical rational disposition concerned with what is
good and bad for human beings” (Aristotle, 1985:273). Prometheus does not seek
to be a redeemer who brings wisdom or justice; rather, he is a mere technical
mediator responding to human deficiencies, embodying the “human wisdom that
contrives to replace the natural fire denied by Zeus with a technical fire”
(Hesiod, 1978:98).
Prometheus
is not a demigod (the result of the union between a deity and a human), nor an
Olympian god, but a Titan—a pure god, preceding the Olympians. Yet, although
the Olympian gods defeat the Titans, Zeus keeps Prometheus close because of his
prophetic capacity, which he wishes to exploit.
The gift of
fire to humans, however, is intolerable to Zeus, who condemns Prometheus to
perpetual suffering: he chains him to a rock, and each day an eagle devours his
liver, which regenerates at night (Hesiod, 1978:93). Heracles (with Zeus’s
consent, according to some versions) eventually puts an end to this suffering
and frees Prometheus—not because he seeks to end the punishment, but because it
ceases to be politically necessary as a demonstration of Zeus’s power and
serves instead to vindicate Heracles’ name among humans—pure politics.
The
punishment inflicted on Prometheus for transmitting fire was not a “moral
sanction” but a penalty for having produced a structural imbalance: fire in
human hands alters the previous equilibrium without introducing a new order. In
Plato, this conception reveals its whole political meaning: because of
Prometheus, technique reaches humans before the virtues of dikē
(justice) and aidōs (moral restraint, the capacity to contain action out
of consideration for others and the community) (Plato, 1981:526–527), which are
essential for its use within social life.
Prometheus as an Expansion of the Myth of Antaeus
From the
Platonic perspective, technique without ethics is self-destructive: while
technique represents the power of a ship’s engine, dikē and aidōs
function as its rudder and compass (or GPS), without which the vessel cannot be
steered or given direction.
When we
read the myth of Prometheus through the lens of Antaeus, we see that the former
describes a transfer of power: the expansion of the potential of fire, which
burns in the absence of restraint and justice—elements as necessary to handling
fire as contact with the ground was for Antaeus.
Contemporary
AI reproduces this pattern with precision. Technical capacity expands at a far
greater speed than our ability to fully understand it, control its potential
harms to society as a whole, and distribute its costs and benefits in a fair
(equitable) manner.
Returning
once again to the classics and applying Aristotelian concepts, AI belongs to
the realm of poiesis, not praxis. That is, it “does things,” but
lacks the capacity to “decide what is right,” because it does not act in human
situations where such judgment is relevant. And although it produces effects,
it does not exercise phronesis: it does not understand the meaning of
what it does and is therefore not imputable in itself. The fact that AI
actively mediates human action does not turn it into an agent. Instead, it
transforms the material and symbolic conditions under which humans act.
From an
Antaean perspective, this means that the Promethean expansion of AI is real,
but not autonomous. If Antaeus embodies absolute dependence on the original
source, Prometheus represents the “human wisdom that contrives to replace
natural fire (…) with a technical fire” without possessing the virtues required
to handle it (Hesiod, 1978:98).
Prometheus
does not contradict Antaeus; he presupposes him and carries him to a higher
level of complexity.
Icarus: Not only a Human Failure
For me,
this is one of the saddest myths, and it is often used as a moral warning;
here, however, it serves as a structural diagnosis.
Ovid
(1980:6–7) tells us that the great architect and inventor Daedalus was held
captive on the island of Crete together with his son Icarus, by order of the
king. Daedalus decides to create two pairs of wings—one for himself and one for
his son—to escape the island-prison.
Daedalus is
fully aware of the design of the wings, their capacities, and their limits, and
he instructs Icarus not to fly too low, lest the water soak the feathers and
make them too heavy, nor too high, lest the sun’s heat melt the wax that binds
them together. And so both begin to rise.
After
becoming accustomed to the vertigo of flight, Icarus gives in to the sensations
of freedom and exhilaration he feels and, ignoring his father’s warnings, flies
too high. The sun melts the wax on his wings and, before Daedalus’s helpless
gaze, Icarus plunges into the sea and dies. His fall is not due to poor design
or lack of information about how to use the artifact, but to having ignored its
limits by altering the contextual conditions of its operation. Icarus does not
fall because of a technical flaw, but because he uses technology beyond the
boundaries that made its operation possible.
Icarus as a Structural Moment of the Myth of Antaeus
From an
Antaean logic, Icarus represents a critical moment: power is now lost neither
through lack of contact with the ground nor as punishment for breaking the
balance of things. Instead, a change emerges that—far from being accidental or
correctable through better individual use—arises when scale and automation
alter the very conditions under which the technical system operates. What takes
place is a structural change in the regime under which the artifact operated,
which also transforms the relationship between technical potential and its
support, because in Icarus, technology exceeds the operator’s capacity for
control.
That allows
us to think about AI failures without resorting to narratives of individual
error or psychological hybris. Icarus was not arrogant, nor did he seek
to play god. Instead, he faced a problem of design, use, and scale that
exceeded his capacities. Icarus did not fall because he was “incompetent,” but
because he behaved in a profoundly human way within an artificial technological
environment.
When a
system is deployed at scale, decision-making is automated, and infrastructure
is transformed, the result is only a change in the conditions that existed
before its deployment—something that Ihde (2015:24) describes as the “complex
structure of technological multistability.”
As
Peter-Paul Verbeek (2005:114) notes, this change occurs through mediations that
redistribute the meaning of actions without turning artifacts into agents
endowed with their own will. Technical objects do not act as “mere neutral
intermediaries between humans and the world, but as mediators: they actively
mediate this relationship.” In fulfilling their functions, artifacts play a
“profound role in what we do, how we perceive and interpret the world, and what
choices and decisions we make” (Verbeek, 2005:vii).
Seen in
this way, Icarus is neither a strict continuation of Antaeus nor a reversal of
Prometheus, but rather a framing of the structural consequence that arises when
the Antaean bond weakens due to the absence of virtues that should have
preceded technique to allow its socially safe deployment.
Governing AI: Governing the Ground and Mediations
Artificial
intelligence is neither a Promethean excess of technological empowerment nor an
individual Icarian error. It is an Antaean form of power: strong as long as it
remains in contact with its material base, and vulnerable when it loses that
contact.
Governing
AI is not about moralizing algorithms, which lack the capacity for praxis
even if they possess poiesis, and do not exercise phronesis.
Governing AI means “governing the ground” of Antaeus, now transformed into
infrastructure, energy, and data. Paradoxical as it may seem, AI does not
decide; it responds, depending on the conditions of its “ground.”
For this
reason, the three myths discussed here can be logically structured as follows:
Antaeus explains conditional power; Prometheus explains its expansion without
an adequate axiological support; and Icarus explains failure—opening, through
myth, a path to rethinking what AI means by exposing its political dimension.
Control over the material conditions of possibility is control over algorithmic
power. The real problem is not what AI can do, but who controls this “human
artifice of the world” that conditions our existence (Arendt, 2003:14).
The central
political problem of power in AI does not lie in achieving greater computing
capacity or more data—which constitute technological power by themselves—but in
the “ground” that sustains them, and is therefore structural.
References
Apolodoro.
(1985). Biblioteca. Biblioteca Clásica Gredos, (85).
Arendt, H, (2003). La condición humana. Paidós
Aristóteles.
(1985). Ética nicomaquea / Ética eudemia (. Biblioteca Clásica Gredos, (89).
Aristóteles.
(1994). Metafísica. Biblioteca Clásica Gredos, (200).
Blumenberg,
H. (1985). Work on myth. MIT Press.
Bratton, B.
H. (2021). La terraformación. Buenos Aires: Caja Negra.
Calvino, I. (2015). Por qué leer los clásicos. Colección Biblioteca Italo Calvino,
(19). Siruela.
Hesíodo.
(1978). Obras y fragmentos. Biblioteca Clásica Gredos, (13).
Ihde, D. (2015). Postfenomenología y tecnociencia:
Conferencias en la Universidad de Pekín. Sello Arsgames
Levi-Strauss, C. (1995). Antropología estructural. Paidós
Ovidio, P. N. (1980). Metamorfosis (Libros VIII–XV).
México: Universidad Nacional Autónoma de México.
Platón.
(1981). Diálogos I. Biblioteca Clásica Gredos, (37)
Verbeek,
P.-P. (2005). What things do: Philosophical reflections on technology,
agency, and design. Pennsylvania State University Press.