AI citation failures reveal something larger than a technical flaw: a civilisation increasingly efficient at reproducing knowledge while losing track of how it became known.
AI citation failures reveal something larger than a technical flaw: a civilisation increasingly efficient at reproducing knowledge while losing track of how it became known.
The question is no longer whether writers use generative technology. It is what remains of the writer after the technology becomes indispensable.
We are building extraordinarily intelligent machines to solve a problem whose governing assumptions remain remarkably primitive.
Tech CEOs mistake being rewarded by the machinery for understanding it. The resulting confusion is being offered as humanity’s only available future.
We built machines to finish our sentences. Now they are helping us finish the world.
We are not outside the world we are trying to understand. We are (all) implicated in its disorders, participants in its reproduction, and responsible for what happens next.
Communication systems can make an interpretation familiar before it has been examined, significant before it has been understood, and consequential before anyone establishes whether it is true.
The AI business model converts new capabilities into dependencies, then turns the resulting risks, maintenance burdens and failures into reasons for further investment.
A rigged game constrains probabilities; it does not determine every outcome.
We have constructed a civilisation in which recognising existential risk has become another reason to continue producing it.
Conflict builds technology, technology reorganises conflict, and somewhere inside the loop every new problem becomes someone’s next investment opportunity.
Education stabilises knowledge long enough to transmit it, even as the machinery through which knowledge is produced and interpreted accelerates the movement of the epistemic attractor itself.