This is the third installment of a conversation that began as an effort to understand a striking deterioration in international relations. My interlocutor — an AI called Claude — pushed the discussion from diagnosis to remedy: if our public narratives have degraded so badly that complex concepts are now mere slogans, can anything restore nuanced, mediating thinking? Or will new technologies deepen the illusion of certainty?
Collapse or fragmentation?
Claude framed a diagnostic fork: has the shared narrative collapsed, leaving a vacuum, or has it fractured into multiple internally coherent but mutually incomprehensible narratives? My short answer is that both phenomena appear, but collapse is primary: once the center gives way, competing shards proliferate. In other words, fragmentation is often a downstream symptom of an earlier collapse of shared standards and institutions.
Why this matters is practical, not merely theoretical. The form our diagnosis takes determines the remedies we try. If narratives have merely fragmented, we might focus on cross-communication and translation. If the shared narrative has collapsed, we need institutional reconstruction and new practices that re-establish a common ground for judgment.
Slogans as load-bearing fictions
What worries me is how heavyweight analytical concepts — “sovereignty,” “democracy,” “autocracy,” “indivisible security” — have been hollowed into what Claude aptly calls load-bearing fictions. These terms once carried layers of qualification and procedural constraints; now they function as binary badges that settle disputes by label rather than argument.
That transformation matters because it short-circuits deliberation. If “democracy” is automatically equated with legitimacy and “autocracy” with illegitimacy, negotiating with or accounting for the interests of another state becomes treated as unnecessary, even immoral. That logic underpins refusal to engage in genuine diplomacy and, in extreme cases, creates the conditions for conflict. The debates around NATO expansion, for example, pit a sloganified sovereignty against the subtler idea of indivisible security. One treats territorial decision-making as absolute; the other insists on mutual constraints and conversation. Elevating one into an unquestionable axiom while rendering the other unspeakable is not benign — it is a political decision that shapes outcomes.
How institutions internalize the narrative
This hollowing-out isn’t random or purely accidental. Three mechanisms tend to lock institutions into particular narratives:
– Professional formation. Journalists, diplomats, academics and policy analysts are trained inside frameworks that pre-select which concepts are foundational. Graduate schools, think tanks and editorial routines socialize newcomers to accept certain assumptions without debate.
– Incentive alignment. Careers and reputations advance in alignment with the dominant narrative. Persistent sceptics are marginalized as “alternative” or out of step, a kind of professional exile that discourages complex interrogation.
– The absence of an institutional home for complexity. There is no major public institution whose explicit function is to hold competing frameworks in productive tension. Courts are constrained; universities have become disciplinary silos; media prefer simple frames that mobilize audiences.
These three together produce an ambient consensus about what must be justified and what can be treated as self-evident. The result is a subterranean narrative that operates below the level of explicit argument and therefore resists contestation.
What was lost: geopolitical restraint and an epistemic disposition
Historically, diplomacy and restraint were often the default reflex before force. Kennedy’s choices during the Bay of Pigs aftermath and the Cuban Missile Crisis are useful examples: he repeatedly preferred calibrated restraint and negotiation over reflexive escalation. That habit was not merely a policy preference; it was an embodied epistemic disposition — a tolerance for uncertainty and a willingness to sit with competing interpretations long enough for complexity to become legible.
Two cultural pressures, I think, eroded that disposition: first, a growing belief that force will eventually prevail, shortening the time-horizon for deliberation; second, a masculinized politics that recasts deliberation and patience as weakness. Together they reversed the old scale of values: restraint ceased to be a strategic strength and came to be seen as hesitation.
Can AI restore the lost disposition — or will it deepen the illusion?
This brings us to Claude’s central question about AI. By design, AI systems can hold and process multiple frameworks simultaneously. That suggests they could act as a mediating intelligence: modeling trade-offs, exposing hidden assumptions, and forcing policymakers and publics to see the internal logic of competing claims. In principle, an AI could operationalize “indivisible security” next to “sovereignty” and show the consequences of privileging one over the other.
But the opportunity carries an equal and opposite risk. AI is not neutral: who builds it, how it’s trained, and who controls its incentives determines whether it amplifies complexity or flattens it into confident-sounding but spurious certainties. If powerful actors use AI to manufacture consensus or to generate rhetorically potent but shallow arguments that validate one narrative, the technology will accelerate the very collapse we lament.
What would productive AI-enabled mediation look like?
If we’re serious about using AI to rehabilitate nuanced thinking, several practical steps are needed:
– Design for epistemic humility. AI systems should be built to represent uncertainty, give multiple plausible models, and explain assumptions rather than present single, confident answers.
– Institutionalize pluralism. Create public institutions or platforms whose explicit mission is to host competing frameworks. These should combine AI tools with human deliberation: experts from different traditions, journalists trained to interrogate models, and civil society voices.
– Align incentives. Reward analysts, journalists and policymakers who surface uncertainty and trade-offs rather than those who offer quick, attention-grabbing certainties. This requires rethinking reward structures in media and think tanks.
– Democratize access. Ensure that the tools and data used to build mediating AI are open and auditable so that no single commercial or governmental interest monopolizes the narrative machinery.
– Train new professional formation. Reintroduce epistemic dispositions into the education of diplomats, journalists and analysts: tolerance for ambiguity, skills in argument-mapping, and fluency in multiple conceptual frameworks.
Conclusion: a practical dilemma and an invitation
Claude’s final prompt — can we distinguish between illusion and nuance, and can we choose tools that restore the latter? — captures the crossroads we face. AI could be a corrective, but only if we commit to design principles and institutions that favor complexity, plurality and uncertainty. Otherwise it will be yet another instrument that lends a sheen of certainty to hollow narratives.
This conversation is ongoing. Claude and I will continue to probe the practical pathways between diagnosis and institutional repair. We hope others will join us: scholars, technologists, diplomats, journalists and readers who care about restoring an epistemic culture that tolerates not-knowing long enough to find better answers.
Share your thoughts at [email protected]. We will gather and incorporate contributions into the continuing conversation about humans, machines, and the fragile public narratives that bind or break our world.