What AI Shows You — and What It Doesn’t
Meike’s Reflections on Artificial Intelligence
Do you prefer to listen to this article? Click below to access our AI-generated audio version!
What AI Shows You — and What It Doesn’t
Meike’s Reflections on Artificial Intelligence
This is the third part of MDI’s Leadership Architect Meike Hinnenberg’s reflection series. You can find parts I and II on our blog page! Stay tuned for more parts to come 🙂
What Leaders See — and What Stays Hidden
As lines of enunciation organize the field of sayability, lines of visibility organize the field of perception. They are conditions of seeing that circulate within a dispositif, determining what can appear as an object, what form something must assume to become perceptible, from which vantage point it is illuminated, and what must recede into shadow for this illumination to hold. A line of visibility is thus a historically specific regime of seeing: a distribution of light and darkness that brings certain realities into presence while casting others into the shadow on which this presence depends.
Michel Foucault traced a transformation of regimes of seeing when he showed how sovereign power, once staged in the blinding spectacle of public punishment, gave way to disciplinary power embedded in architectures of continuous observation. What changed was not only the exercise of power, but the arrangement of the visible itself: spectacle yielded to surveillance, and visibility ceased to be an event and became an environment.
When we turn to the dispositif of Artificial Intelligence, how is the terrain of perception arranged, and which lines of visibility organize this regime of seeing?
How AI Presents Itself: Four Lines of Visibility
One line runs along the interface. Here, Artificial Intelligence appears as responsiveness without delay: dashboards refresh in real time, prompts yield fluent replies, and systems demonstrate competence in carefully staged demonstrations. Intelligence presents itself as performance – immediate, seamless, self-contained. What this line establishes is the perceptible surface of operation: output as event, response as evidence. The system comes into view precisely where it answers.
Line 2: Abstraction — Structure Without Weight
A second line follows the path of abstraction. Models are described by architectures, parameters, and accuracy scores; performance is reported numerically, and improvement is recorded as optimization. Intelligence becomes legible as a formal property, detached from situation and substrate. What comes into view is structure without weight, reasoning without environment, cognition without bodies.
Line 3: Scale — Expansion Beyond Intervention
A third line unfolds at the scale level. Artificial Intelligence appears as planetary infrastructure: billions of parameters, global deployment, continuous operation across time zones and continents. Its magnitude exceeds ordinary perception. Scale produces its own regime of visibility: what emerges is inevitability, momentum – expansion beyond intervention.
Line 4: Neutrality — When Calculation Replaces Judgment
A fourth line organizes neutrality. Artificial Intelligence appears as objective and data-driven. Its operations present themselves as technical processes rather than situated decisions. Judgment appears as calculation; outcomes appear as results rather than interventions. What appears is a world cleansed of politics, in which a large part of responsibility is shifted to the system, and context is leveled out. Neutrality here is not simply descriptive; it is productive, structuring perception so that harm, choice, and embedded values recede into shadow, while the surface of computation shines as transparent and self-evident.
The Illusion of Autonomy — and What It Conceals
Together, these lines compose a regime of seeing in which Artificial Intelligence presents itself as autonomous, immaterial, and inevitable. What appears is intelligence without remainder. Yet regimes of visibility do not simply reveal; they arrange revelation. They produce perceptibility by structuring what cannot be seen at the same time.
By citing Amazon’s crowd-working platform “Mechanical Turk” and recalling its historical namesake – the ostensibly chess-playing automaton constructed by Wolfgang von Kempelen in 1769 – Kate Crawford traces such a line of visibility and its fracture at once. The figure of the seemingly chess-playing automaton, dressed in Ottoman robes and seated before a wooden cabinet topped with a chessboard, appeared to deliberate and decide on its own. When its doors were opened, intricate gears and clockwork were revealed, offering the reassuring image of mechanical reason. Yet this visibility was carefully staged: concealed within the cabinet, a human operator followed the game in darkness, shifting position as panels were displayed to sustain the illusion. What appeared to be autonomous intelligence was, in fact, the surface effect of a hidden human presence.
In recalling this machine, Crawford renders perceptible a continuity that the contemporary name Artificial Intelligence works to obscure: the appearance of autonomy sustained by distributed, hidden work. That Amazon names its global digital labor platform after this deceptive automaton – an illusion built not only on concealment but on the orientalist staging of a racialized figure – is at once cynical and involuntarily revealing. The name preserves, like a fossil in language, a longer history in which intelligence appears at the surface while the labor that sustains it is displaced elsewhere, often across colonial and postcolonial geographies, into bodies that remain structurally unrecognized.
By shifting the vantage point, she intervenes in the regime of seeing itself. What appeared seamless reveals fracture lines; what appeared autonomous reveals dependence. The interface no longer appears as an origin but as a surface.
Behind the Surface: Labor, Matter, and Geography
Behind the abstraction of the model, material infrastructures come into view. Data centers operate at an industrial scale, consuming vast quantities of electricity and water to sustain continuous computation. Their servers depend on the conflict minerals tin, tantalum, tungsten, gold, and rare earth elements extracted from landscapes marked by toxic residues and ecological exhaustion. The expansion of machine learning contributes to growing streams of electronic waste, measured in millions of tons. What appears as immaterial intelligence is inseparable from extraction, depletion, and heat.
Behind the neutrality of data, processes of selection and classification emerge. Machine learning systems depend on vast datasets assembled through human activity: images segmented, sentences evaluated, gestures annotated. Millions of crowd-workers across the world perform these tasks, often for minimal compensation, clicking through thousands of items in repetitive sequences that train systems to see. Content moderators encounter violence, pornography, and degradation so that others encounter sanitized outputs. Their perception becomes part of the system’s sensory apparatus, even as their presence disappears from its representation.
Behind the scale of the system, a geography becomes perceptible: supply chains stretching across continents, data centers situated near sources of energy and water, labor distributed across time zones, extraction zones, and processing facilities linked in continuous operation. What appears as a unified technical object reveals itself as a convergence of environments, infrastructures, and bodies.
Seeing Otherwise: From Output to System
Artificial Intelligence does not simply appear differently once these conditions are seen. The regime of visibility itself is exposed as constructed. The lines that once produced the appearance of autonomy are revealed as arrangements that separate surface from substrate, output from labor, intelligence from matter.
To follow these fracture lines is not merely to see more, but to see otherwise. Intelligence no longer appears as an isolated technical achievement, but as the visible surface of relations extending downward into the earth, outward across the planet, and inward into the perceptual and cognitive labor of others. What had appeared as a self-contained system becomes perceptible as a dispositif: an arrangement that produces both the object and the subjects who sustain it, while organizing the conditions under which this production can be seen or remain unseen.
Meike Hinnenberg
Senior Leadership Architect
Meike Hinnenberg is a trainer and Senior Leadership Architect at MDI Management Development GmbH and specializes in communication, conflict management, diversity & inclusion, and lateral leadership.
Join us on May 5th for our yearly Leadership Horizon Conference!
