Rationalism has two faces. One is luminous: the discipline that broke superstition’s monopoly, built modern science, stabilised engineering, made medicine reliable, and turned fragile lives into survivable ones. Rationalism, at its best, is not arrogance. It is a refusal to be governed by moods, priests, or crowd contagion. It is the insistence that claims must answer to constraints.

The other face is darker: rationalism as a civilisational control technique — a way to compress complexity into procedures, metrics, and “reasonable” narratives that make coordination easy while quietly liquidating judgement.

Both faces are real. The problem is not that rationalism exists. The problem is that rationalism has been promoted into a religion precisely at the moment when its most useful function—complexity reduction for interdependence—turns into a liability.

Rationalism’s forgotten origin: coordination under pressure

Rationality did not win history because it was “truer"; it won because it was operational. Large-scale societies require shared protocols: legal forms, accounting systems, standardised education, bureaucratic decision paths, and repeatable explanations. Rationalism became the lingua franca for this: if reasons can be stated, evaluated, and reproduced, systems can scale. A civilisation of interdependence needs a common format that compresses lived ambiguity into transferable decisions.

In that sense, what is commonly called “rational” is often not the pinnacle of insight. It is the shortcut logic of interdependence: a way to reduce complexity in order to enable external steering—coordination, compliance, prediction, and allocation.

That is why rationalism is so often attached to optimisation. Optimisation is the dream of interdependence: if a single “best” path exists, the swarm can move without hesitation.

But this reveals the hidden trade: rationalism frequently buys coordination by selling off interior reality.

Where rationalism is unbeatable

A differentiated critique must admit this: rationalism is not merely ideology. It is a genuine epistemic advance in specific domains.

  • When the environment is stable enough that models can converge.

  • When the goal is specifiable (build a bridge, reduce infection rates, detect fraud).

  • When errors are measurable and learning loops are tight.

  • When claims must be falsifiable rather than merely persuasive.

In these regimes, rationalist discipline is mercy. It prevents magical thinking from becoming policy. It prevents charisma from substituting for evidence. It builds institutional memory. It reduces the cost of repeating the same mistakes.

The problem begins when rationalism is treated as the supreme frame for all frames.

The rationalist error: confusing coherence with calculation

Rationalism often claims "coherence". But coherence has two species:

  • Computational coherence: internal consistency in a model or argument.

  • Infosomatic coherence (human coherence): integration across perception, consequence-sensing, values, and lived accountability.

The first can exist without the second.

A perfectly consistent argument can be perfectly dead.

A model can be elegant and predictive and still be mis-specified at the level that matters: what counts as a relevant variable, what is being optimised, and what is being destroyed while the KPI improves.

Rationalism excels at refining answers inside a frame. It is far weaker at answering the more dangerous question: Who selected the frame — and why?

The moment a civilisation confuses calculation with coherence, it becomes easy to govern by plausibility rather than responsibility.

The shadow function: rationalism as an alibi

Here rationalism becomes brutal.

In many institutions, “rationality” is not used to think. It is used to avoid thinking.

  • Procedure becomes an alibi for choice.

  • Metrics become an alibi for judgement.

  • Consensus becomes an alibi for truth.

  • Expertise becomes an alibi for accountability.

This is not a moral accusation. It is a systemic pattern. Interdependence rewards anything that produces legible alignment. Rationalism is legibility. And legibility can be manufactured.

So rationalism becomes a theatre in which institutions perform responsibility while outsourcing the burden of consequence-bearing decisions.

That is why modern systems can be “rational” and still behave pathologically: the rationality is real, but it is rationality in service of a mis-specified survival loop.

Why AI exposes rationalism’s limits

AI does not merely automate tasks. It commoditises the visible layer of rationalism: fluent reasoning, consistent argumentation, plausible explanation, and rapid synthesis.

This is the scandal: much of what passed as “being rational” in professional life was never orientation — it was the ability to produce convincing symbolic artefacts.

AI can now do that at scale.

So the bottleneck relocates upward: from producing reasons to holding criteria. From generating coherence to selecting what coherence is for. From optimisation to stop-conditions.

In other words: AI reveals that rationalism—when treated as output rather than responsibility—was always a substitute currency.

The counterpoint: anti-rationalism is not the cure

At this point, a cheap critique would declare rationalism “dead” and celebrate intuition, spirituality, or aesthetic knowing.

That would be an error of the same type: swapping one total frame for another. Because the opposite of rationalism is not depth. It is often impulse dressed as authenticity. The question is not “rationality or not". The question is, what level of human functioning is being protected?

A mature civilisation needs rational methods inside domains where they belong—yet must also cultivate capacities that rationalism cannot replace:

  • Judgement under uncertainty.

  • Criteria formation when goals are contested.

  • Responsibility that does not evaporate into procedure.

  • The ability to refuse acceleration.

  • Orientation when no model converges.

This is the domain of what can be called epistemic integrity: not correctness, but consequence-bearing coherence.

From rationalism to orientation

Rationalism asks, 'What is the best answer?'

Orientation asks, 'What is the right question — and what is the cost of asking it?'

Rationalism seeks to optimise within a given landscape.

Orientation asks whether the landscape itself is the trap.

This is why the contemporary crisis is not informational. It is criterial. Not a shortage of knowledge, but a shortage of subjects capable of holding criteria without collapsing into swarm dynamics.

Rationalism was historically a tool that made interdependence workable. But interdependence, left unchecked, produces a world of role-players and ego-constancies—actors stabilised by approval loops rather than by inner decision capacity. In such a world, rationalism becomes an instrument of control: it standardises minds for coordination.

The next step is not “more rationality".

It is the emergence of subject autonomy in becoming: the capacity to remain decision-capable without hiding behind consensus, metrics, or institutional scripts.

A differentiated thesis

So the critique resolves into a clean proposition:

  • Rationalism is indispensable for building reliable systems in stable domains.

  • Rationalism is insufficient for governing reality when stakes are irreducible, values collide, and the frame must be chosen.

  • Rationalism becomes dangerous when it functions as an alibi for avoiding judgement and dispersing responsibility.

  • AI accelerates the collapse of rationalism-as-performance by making rational-sounding output cheap.

  • The scarce resource is now orientation: criteria, stop-conditions, accountability, refusal, and consequence-bearing judgement.

This is not an attack on reason. It is a demand that reason stop impersonating sovereignty.

The exit from the rationalist trap

The rationalist trap is not “thinking too much". It is thinking in a way that remains compatible with external steering. The exit is not irrationality. The exit is non-reductive coherence: an architecture where the subject can hold criteria under complexity without needing the swarm to provide identity.

A civilisation capable of that shift would still use rational methods—but would no longer worship them. It would treat rationality as a tool for local problems while building its legitimacy on a higher capability: orientation that remains intact under acceleration.

That is the real transition.

Not from humans to machines.

From rationalism as a social ritual to orientation as a civilisational condition.