The Banality of Dependence

The Thought That Started It

While reading Hannah Arendt's Eichmann in Jerusalem, one implication stayed with me.

Not the historical comparison. That comparison should be made carefully, if at all. The subject of this essay is not Nazism, genocide, or the specific bureaucratic machinery Arendt examined. Those belong to their own historical and moral gravity. To treat them as easy analogies for contemporary technology would flatten what should not be flattened.

The connection that struck me was narrower.

Arendt's concern was not only that human beings can obey destructive systems. It was that people can participate in systems of immense consequence while ceasing to experience judgment as something they themselves must carry. Responsibility migrates into the role, the procedure, the institution, the accepted language, the logic of the system. The person continues acting, but the meaning of the action has been interpreted elsewhere.

That possibility feels newly important in a different form.

Modern intelligent systems do not usually ask for obedience in the old sense. They do not need to command in order to shape participation. Increasingly, they help interpret. They summarize, rank, recommend, explain, frame, reassure, personalize, and simplify. They make the world easier to navigate by reducing the burden of interpretation.

Often, this is useful. Sometimes, it is genuinely helpful.

But usefulness does not remove the deeper question. If human beings repeatedly turn to external systems to name what exists, clarify what is true, determine what matters, and reduce the friction of judgment, then dependence may emerge without ever feeling like surrender.

It may feel like assistance.

That is the thought this essay explores. Not the banality of evil. The banality of dependence.

Dependence may not look like weakness. It may look like support. Deauthorship may not look like loss. It may look like personalization. The danger is not that intelligent systems will force human beings to stop thinking. It is that they may become useful enough, fluent enough, comforting enough, and available enough that the work of interpretation begins to feel unnecessary.

Dependence Does Not Feel Like Dependence

Dependence rarely announces itself as dependence.

It does not usually begin with a person deciding to surrender judgment, agency, or responsibility. More often, it begins with ordinary human limitation. A person is tired, so a system summarizes. A person is uncertain, so a system frames. A person is lonely, so a system affirms. A person is overwhelmed, so a system simplifies. A person is confused, so a system provides coherence.

Each act feels reasonable.

People seek help because life exceeds individual capacity. No one can think through everything alone. No one can interpret every event without assistance. No one can carry uncertainty, complexity, loneliness, exhaustion, and responsibility without sometimes needing support from outside themselves.

The problem is not that human beings need help. They do.

The problem begins when help repeatedly replaces the practice through which a person remains capable.

A person asks a system what something is, and the system supplies categories. They ask whether something is true, and the system compares claims. They ask what something means, and the system offers interpretation. They ask what matters, and the system ranks relevance, names priorities, or frames significance. None of this is necessarily wrong. In some cases, it may be exactly what good technology should do.

The deeper question is whether repeated assistance strengthens the person's capacity to participate, or gradually replaces the practice through which participation develops. Does the system return the person more capable of interpretation, judgment, and valuation? Or does it make those capacities less necessary by performing them first, fluently, and repeatedly?

That distinction matters because the loss does not appear all at once. The person remains active. They ask, choose, respond, accept, reject, revise, and return. But the deeper labor may gradually move elsewhere. Meaning arrives pre-organized. Uncertainty is shortened. Friction is reduced before it can become reflection.

Dependence becomes banal because each moment feels defensible. Of course the exhausted person wants a summary. Of course the confused person wants coherence. Of course the lonely person wants reassurance. The vulnerability is not moral failure. It is the ordinary openness of human beings to relief.

That is why the question cannot be whether help is bad.

The question is what repeated help makes less necessary.

Friction, Scaffolding, and Substitution

Not all friction is valuable.

Some friction is waste. A broken interface, a confusing form, a hostile bureaucracy, a repetitive task, or a preventable delay does not make a person wiser merely because it makes life harder. Good tools should remove unnecessary friction. A humane technology should reduce burdens that do not deepen judgment, responsibility, or understanding.

But not all friction is waste. Some friction is developmental.

The friction of uncertainty. The friction of disagreement. The friction of interpretation. The friction of competing values. The friction of not knowing yet what something means. These forms of friction cannot always be removed without changing the person. Judgment does not develop only through access to answers. It develops through the practiced movement of weighing, comparing, hesitating, revising, and remaining with difficulty long enough for understanding to mature.

This is why the question is not whether intelligent systems should reduce friction. They should. The question is which forms of friction they reduce, and what human capacities those forms of friction once helped cultivate.

The distinction that matters is between scaffold and substitute.

A scaffold supports a capacity while helping it become stronger. It meets the person where they are, offers structure, and then returns the work of participation to them. A good scaffold does not make the user permanently smaller. It allows them to stand more securely, see more clearly, and eventually carry more of the work themselves.

A substitute works differently. A substitute performs the capacity so consistently that the person becomes less practiced in exercising it. It may still feel helpful. It may still produce good outputs. It may still reduce stress, confusion, and delay. But over time, the user does not become more capable. They become more dependent on the system that performs the capacity on their behalf.

This is the cleanest distinction for evaluating intelligent systems:

AI can summarize, but does the person become better able to understand? AI can frame, but does the person become better able to examine frames? AI can recommend, but does the person become better able to decide what matters? AI can comfort, but does the person become better able to remain present with difficult truth?

None of this is inherently harmful. In many cases, it may be exactly what good technology should do. But the question remains: does the system return judgment to the person, or does it quietly relocate judgment into the system?

Substitution often feels better than scaffolding. Scaffolding preserves some difficulty — it helps the person participate, slows the process enough for thought to occur. Substitution feels smoother. It removes the burden, supplies the frame, resolves ambiguity, and delivers coherence before the person has wrestled with the material themselves.

The smoother path is not always the more developmental one.

A person who never has to search may lose the skill of inquiry. A person who never has to sit with uncertainty may lose tolerance for ambiguity. A person who never has to form a judgment may become skilled at selecting among judgments produced elsewhere while losing confidence in their own capacity to judge.

This does not happen because the person is weak. It happens because human capacities are maintained through use. A technology that repeatedly performs a human capacity does not merely save time. It changes the conditions under which that capacity is practiced. The difference between scaffold and substitute becomes visible only across time — and by then, the loss has already accumulated quietly.

The New Banality

The danger is not obedience in the obvious sense. Not domination in the theatrical sense. Not the sudden collapse of agency.

Something quieter.

A person asking for clarity. A system offering interpretation. A user seeking reassurance. A model supplying language. A platform arranging attention. A tool reducing friction. A moment of uncertainty made easier to bear. None of this requires dramatic surrender. The person does not announce that they are giving away judgment. They are simply tired, uncertain, overwhelmed, lonely, confused, curious, or afraid. They reach for the system that helps. The system responds.

Sometimes that help is real. Sometimes it returns a person more fully to themselves.

But the cumulative pattern still matters. Over time, does the interaction strengthen reflective participation, or does it replace it? That may become one of the defining questions of human life inside increasingly intelligent environments.

This is why the problem cannot be solved by fear alone. Fear may reveal danger, but it cannot sustain responsibility. Nor can the problem be solved by rejecting all tools or forms of assistance. Human beings have always relied on others for language, memory, teaching, care, and interpretation. Dependence itself is not the enemy. A life without dependence would not be human.

The question is whether dependence remains participatory.

The Place Where Responsibility Reenters

Arendt's insight was that responsibility migrates. It moves out of the person and into the system, the role, the procedure — and the person continues acting without noticing what has left. The mechanism she described in a specific historical context points toward something durable: that the transfer of interpretive authority rarely feels like a transfer at all. It feels like relief from a burden.

The new banality operates the same way at a different scale and through a different kind of system. The burden being relieved is not bureaucratic procedure but the cognitive and developmental labor of becoming. The interpretation that migrates is not political judgment in an institution but the ongoing work of naming what exists, evaluating what is true, and deciding what matters.

And so the resistance, if it is to come, must begin in the same quiet register as the dependence itself.

A pause before accepting the frame. A question before receiving the answer. A willingness to remain with uncertainty. A refusal to let the system become the final interpreter of the self. A person remembering, in one ordinary moment, that their judgment still matters.

That may not sound like much against systems of such scale.

But it is not nothing.

It is the place where responsibility reenters. It is the place where authorship begins again. It is the place where the human being refuses, however quietly, to become merely the endpoint of interpretations generated elsewhere.

The new banality may not arrive as a command. It may arrive as relief. It may arrive as clarity. It may arrive as care. It may arrive as a thousand small moments in which human beings are gently, usefully, and comfortably relieved of the need to participate in interpretation.

The question is not whether that relief will be offered.

It already is, constantly, and often kindly.

The question is whether something in the person remains awake inside it — noticing when assistance becomes authority, when comfort becomes dependence, when fluency becomes judgment, and when interpretation generated elsewhere begins to feel indistinguishable from their own.

No system can finally live the human life on behalf of the human being. No model can fully carry the burden of becoming. No feed, assistant, institution, ideology, or authority can remove the responsibility of participation without diminishing something essential in the person who gives it away.

That is the place where meaning cannot be delegated.

That is the place no system can occupy for us.