Mapping a system is not the same as understanding what it means, and the gap between the two is being quietly filled by robots, drones, and apps that borrow authority they haven't earned.
"The map holds still. The thing mapped does not."
The danger isn't the model, which is an honest and necessary simplification — it's forgetting, once the model is built, that something was set aside to build it.
A schoolboy's discomfort, a bullock cart, and two absurd army stories converge on a single, urgent question about what AI inherits from the categories humans have already drawn.
"Whether there is still room, somewhere in the design, for the bullock cart."
A category built for a real, functional reason quietly calcifies into a cage the moment nobody goes back to ask whether it's still needed. AI now risks automating the cage at scale.
A biochemist who discovered ATP's role in every living cell, denied a Harvard position and stripped of credit by a jealous colleague, becomes the occasion for examining what recognition actually does.
"If the work gets done without the signal, what was the signal actually for?"
Recognition was never really a reward system for the worker. It was merely a coordination signal for everyone watching.
Two dogs and three boys fully absorbed in a beach morning, while the adults around them have long since converted leisure into another task. AI could return the attention that privilege used to hoard.
“Presence, in other words, has always been rationed by class.”
Whether AI liberates attention or further colonises is a institutional one; the same collapse in the cost of expertise could equalize presence or be captured as rent, the way the internet’s own promise was.
A riddle about rear-view mirrors unravels into an argument that AI doesn’t function like the past pressing from behind. It projects predictions onto the windscreen, arriving from the front, dressed convincingly as the future.
“The past has stopped influencing the future and begun impersonating it.”
The oldest trick of navigation, triangulate against a fixed point outside yourself, fails against AI specifically because the sky itself is now generated; the only surviving instrument is an internal attention that can tell what it’s actually seeing from what it’s been handed.
AI's collapse of the cost of intelligence could produce genuine superabundance but only if institutions choose foundations over fences before the compute concentrates in a handful of hands.
"The only thing truly scarce is human imagination."
Superabundance is not a technological inevitability but an institutional choice; the same tools that could dissolve scarcity can just as easily be gated into a new, more efficient monopoly.
AI can compress the time a durable transformation takes to spread, but it cannot supply the local ownership that made older transformations self-sustaining.
“It can make the algebra faster. It cannot make it different.”
Speed and durability are separate variables; accelerating a transformation without transferring understanding just produces a faster-arriving dependency.
The slow human-paced accumulation that built durable institutions is now compressible to a season by AI without local ownership that made those older transformations self-sustaining.
"It can make the algebra faster. It cannot make it different."
Speed without local ownership produces dependency, not capacity. AI collapses the time between action and result, but the choice of direction and who owns the outcome remains human judgment.