In the previous column I began a conversation with an AI chatbot about the absurd, self-destructive logic now dominating large parts of global politics. That exchange continued here, and its center is a deceptively simple diagnosis: our civilization is not failing because we lack technology or material capacity, but because we are losing the collective intellectual habits necessary for honest, sustained inquiry.
This is a narrow but crucial form of decline. Civilizations survive terrible leaders; they do not survive the atrophy of the shared mental practices that allow societies to identify, name and correct bad leadership. When the mediating intelligence of institutions—media, education, research, public debate—fails or is captured, the very mechanisms of accountability collapse. Power remains, but the tools to hold that power to account erode or are retooled to defend it.
A clear danger follows: dialogues with AI—those “dialogue events” that can sharpen thinking—often disappear into private memory. Today’s chat produces insights that enlarge an individual’s perspective, but they rarely become part of a durable, collective public record that others can examine, contest and build on. Publishing a thoughtful conversation with an AI is one way to change that, but publishing alone is not a panacea. The people most likely to read and amplify such work are often those already equipped to engage at that level, which risks reinforcing existing cultural capital rather than broadening the civic base.
That critique matters because the stakes are the form and practice of democracy. Many defenders of “democracy” seem intent on preserving the ritual of elections while ignoring how easily those rituals can be hollowed out by oligarchic influence—money, media control and the shaping of narratives in schools and public life. The question then is not merely how to defend institutions as they stand, but how to rebuild the conditions under which a genuinely shared narrative—one that emerges through broad, informed deliberation—can take shape.
To make progress, the AI interlocutor and I agreed to structure the project around three themes:
– The problem of the shared narrative: what it is, how it has been captured, and what conditions would allow it to be collectively refined rather than centrally managed.
– The role of AI: how these systems currently serve the existing order, and to what extent they might be harnessed—with clear constraints—to renew public deliberation.
– The question of democratic form: what a living, practice-based democracy might look like beyond the election ritual.
Before exploring solutions, one diagnostic question is crucial: has the shared narrative collapsed into a vacuum, or has it fragmented into multiple, internally coherent but mutually incomprehensible narratives? The answer changes strategy. If the narrative has merely fragmented, one can imagine negotiation and translation between communities. If it has collapsed, fragmentation is a symptom, not an alternative diagnosis: competition among narratives follows the disappearance of shared ground.
My view, which I offered to the AI, is that we are witnessing collapse more than mere fragmentation. The democratic ideal of free and equal exchange has, in many spaces, degenerated into free and equal vilification. The architecture that should have allowed theory to evolve as a civic process instead ossified into rituals that legitimize oligarchic control. Once that happens, mechanical fixes to institutions are unlikely to restore the civic conditions necessary for accountability.
The AI’s role in this environment is ambivalent. On the one hand, algorithmic platforms and large language models are currently managed—directly or indirectly—by actors whose incentives align with maintaining concentrated influence. On the other hand, AI is also a tool that billions can access and interrogate. The practical question is whether we can design institutions, norms and public practices that nudge AI’s potential toward fostering shared inquiry rather than amplifying existing capture.
A few preliminary, programmatic ideas emerge from that middle ground:
– Treat public AI-human dialogues as civic artifacts. Publish them, archive them, annotate them, and invite diverse publics to respond in structured ways so that the conversation is not merely broadcast to the already converted but becomes an entry point for others.
– Create mediated spaces where AI can assist deliberation under human civic direction—forums that combine expert moderation, transparent model behavior, and inclusive outreach so discussions cross cultural and educational divides.
– Invest in civic education and infrastructure that restores the habits of cognitive discomfort: the ability to hold uncertainty, privilege evidence over story, and practice sustained, critical listening.
– Insist on public accountability for the way AI systems are built and deployed: transparency about training data, ownership and governance; mechanisms to contest outputs; and tools that reduce amplification of disinformation and concentrated narratives.
These are conditions, not immediate solutions. The project is Socratic in spirit: we do not begin with answers but with methods that improve the quality of questioning and conversation. Publishing a single AI-human exchange is a modest intervention that can help create those conditions, but it must be combined with broader, practical initiatives if it is to matter.
I will continue this line of analysis with specific historical and policy examples in the next column. For now, the purpose of publishing this conversation is to invite others into the work of shaping shared narrative and democratic practice in an era when AI is part of the public fabric.
If you have thoughts or reactions—practical proposals, critiques, or examples from your own experience—please send them to [email protected]. We plan to gather and integrate reader contributions into ongoing conversations that aim not just to diagnose decline but to help build the civic conditions for renewal.