The Reluctance to Think Multilaterally

The world is multilateral. Most thinking tools people use are bilateral. The result is a systematic simplification of reality that does not arise from lack of intelligence — but from structures that reward bilateral thinking and penalise multilateral thinking. Artificial intelligence is the first tool that can structurally change this asymmetry. Whether it does so does not depend on the technology.

I. What Multilateral Thinking Is — and Why It Is Rare

Bilateral thinking operates on a single axis: right or wrong, safe or risky, left or right, friend or foe. It is fast, decidable, communicable. It generates clear positions that can be defended. It is the thinking that institutions need in order to function — because institutions need decisions, and decisions need simplification.

Multilateral thinking operates simultaneously on several axes. It holds contradictory truths side by side without resolving them. It does not ask "is this right or wrong?" but "under which conditions is this right, under which wrong, and which conditions currently apply?" It is slow, difficult to communicate, and generates positions that are hard to defend — because they contain reservations.

In a world where career depends on visibility, visibility on clarity, and clarity on simplification, multilateral thinking is a structural disadvantage. Those who say in a meeting "this is more complicated than it seems, and the answer depends on several factors we do not yet fully understand" — are right and lose. Those who say "this works, let's do it" — may be wrong and win regardless.

Bilateral thinking is not stupidity. It is the rational adaptation to systems that reward decisiveness and penalise nuance. Multilateral thinking is not cleverness. It is a disposition that can survive in certain structures — and cannot in most.

II. The Forms of Bilateral Thinking

The handbook engineer thinks bilaterally: what is in the manual exists. What is not in it does not exist. The axis reads: norm versus deviation. This is not a personal weakness — it is the rational response to a system that penalises deviation from norms and rewards conformity. Those who defend the norm are safe. Those who deviate from it bear the risk alone.

The politician thinks bilaterally: left versus right, government versus opposition, friend versus foe. Political communication enforces this simplification — not because politicians are less intelligent than others, but because a differentiated position does not survive in public. The sentence "this is a complex question with arguments on both sides" is politically useless. The sentence "we have the solution and the others do not" is politically effective — and empirically almost always wrong.

The manager thinks bilaterally: risk versus safety, investment versus cost-saving, innovation versus stability. Quarterly logic enforces bilateral decisions because it shortens time horizons. Those who think multilaterally — who ask how a decision will affect three different system levels in five years — are too slow for the reporting cycle.

And the citizen thinks bilaterally: because the media that deliver his information are bilaterally structured. Because the social networks in which he communicates reward bilateral reactions — outrage or approval, sharing or ignoring. Because the political culture in which he lives demands bilateral assignments. One must be for something or against something. The both-and is suspect.

III. What Was Missing Until Now

There have always been people who could think multilaterally. They are recognisable by a specific characteristic: they are uncomfortable. Not because they are difficult, but because their answers provide no simple instructions for action. The multilaterally thinking adviser who says "it depends" is less in demand than the bilaterally thinking adviser who says "do this." That the former is usually right and the latter usually oversimplifies changes nothing about demand.

The tool problem was always the same: those who want to think multilaterally need access to several perspectives simultaneously. They need the engineer, the economist, the historian, the sceptic — in the same room, at the same time, on the same question. This was organisationally costly, economically inefficient, and humanly problematic: people bring their positions, their career interests, their NIbyM, their silence cartel. The historian who tells the engineer "we have tried this before, and it did not work because..." — is not heard in most meetings, because he slows the meeting down.

Libraries, databases, search engines did not solve the problem — they made information accessible, but not perspectives. A search engine gives no answers, it gives documents. The synthesising, the weighing, the holding of contradictory truths — that remained a human task. And humans rarely do it well, because the structures in which they work do not encourage it.

IV. What AI Changes Structurally

An AI has no NIbyM. It has no career that depends on the decision. It belongs to no silence cartel. It has no position in the three-role triangle to defend. It can simultaneously play the engineer and the sceptic, the historian and the economist, the advocate and the critic — without any of these roles corresponding to an identity that needs protecting.

That is not a small difference. It is a structural break with everything that was previously possible. For the first time, a single person with a question can receive several serious perspectives on that question simultaneously — without conference, without budget, without the social costs of contradiction. The AI can say: "From an engineering perspective, much speaks in favour. From an economic history perspective, there are three counterexamples you should know. From the perspective of your company's risk structure, the decisive question is a different one." A human interlocutor could say the same — but rarely does, because he pays a price for it.

The possibility of multilateral thinking has always existed — as a capacity, as a disposition, as an intellectual ideal. What was missing was a tool that makes this possibility accessible to everyone who wants to use it. Not for the few who can afford an advisory board. Not for the institutions that can fund interdisciplinary teams. For anyone who has a question and is willing to receive a complex answer.

V. The Limit of Technology

Here the good news ends — or more precisely: here the actual question begins.

An AI used as a confirmation machine changes nothing. Those who ask it "confirm that my decision is correct" — receive bilateral thinking with AI effort. Those who ask "what arguments support my position?" — receive an elaborated version of their own conviction. The technology does not change the quality of the answer if the quality of the question is poor.

Multilateral thinking with AI requires a willingness that the technology cannot generate: the willingness to be contradicted. Those who ask an AI and then select the answers that support their existing position — have used a new technology for an old reflex. Confirmation bias does not disappear through AI. It merely finds new instruments.

The actual competence that enables multilateral thinking with AI is not technical. It is the competence to frame a question in a way that invites contradiction. It is the willingness to receive an answer that complicates rather than confirms one's position. It is the ability to live with the discomfort that arises when one recognises that reality is more multidimensional than the model with which one has previously described it.

AI makes multilateral thinking possible. It does not make it automatic. The tool exists. The willingness to use it is a human decision — and it is exactly as rare as before.

VI. Who Uses It — and Who Does Not

Early observations are sobering and encouraging in equal measure. Sobering: a large part of AI usage is bilateral. One asks to produce texts, not to expand perspectives. One asks to become faster, not to think more deeply. One asks to complete tasks, not to open questions. AI is used as an efficiency tool — which it also is, but not only.

Encouraging: there are people who use it differently. Who present a decision they have already made to an AI — not to receive confirmation, but to hear the counterarguments they have overlooked. Who illuminate a technical solution they have developed from the perspective of the historian, the economist, the lawyer. Who interrogate a political judgement they considered secure for its premises. That is multilateral thinking with AI. It is possible. It is happening.

The difference between the two groups is not intelligence. It is the disposition described in the falsification essay: the willingness to take what one sees seriously — even when it contradicts what one would prefer to believe. This disposition was always rare. AI has not made it more frequent. But it has given it a tool that multiplies its potential.

VII. The Institutional Question

The actually interesting question is not the individual one. It is the institutional one: what happens when institutions — companies, authorities, political parties — begin to use AI for multilateral thinking? When the decision structure that enforces bilateral thinking meets a tool that delivers multilateral perspectives?

The pessimistic answer: institutions will configure AI to produce bilateral answers. They will build systems that confirm existing decision structures rather than challenge them. They will filter out the complexity that AI can deliver and retain the clarity that their processes require.

The optimistic answer: some institutions will recognise that multilateral thinking is a competitive advantage. That the organisation making decisions on the basis of several simultaneously held perspectives systematically makes better decisions than the organisation deciding bilaterally. That the price of complexity is lower than the price of false simplification.

Both answers will prove simultaneously true — in different institutions, different cultures, different industries. Selection pressure will decide which prevails. That is not a technical question. It is a question of the structures that reward decision quality — or do not.

Bilateral thinking led Europe into the dead ends described throughout this essay series. The energy dependency was the result of bilateral weighing: cheap in the short term versus risky in the long term — and the short-term side won. The defence incapacity was the result of bilateral weighing: spending today versus security tomorrow — and spending lost. Multilateral thinking would have framed both questions differently. The tool for doing so now exists. The willingness to use it is the open question.