Before the railways, time was local.
Each town set its clocks by the sun. Noon in Bristol was not quite noon in London, and that didn’t matter because nothing moved fast enough between them for the difference to cause trouble. Time was contextual. Embedded. Agreed through proximity.
Then the railways arrived and distance began to collapse. What had been a harmless variation became an operational risk. A train could leave on time in London and arrive late in Bristol without slowing down. The problem wasn’t mechanical, it was conceptual. Different places were operating on different assumptions about reality.
Railway companies standardised time. Not because they had collectively agreed on a new philosophy of temporality. Because coordination demanded it.
Alignment came before agreement.
Most organisations run quite happily on local time. Different teams hold slightly different definitions of risk, of done, of acceptable error. These differences are rarely written down. They are absorbed, negotiated, smoothed over through experience and proximity. When work moves at human speed, tacit coordination is usually enough.
AI changes the speed and the connectivity.
It asks for clearer definitions. It exposes where thresholds were assumed rather than agreed. It has no tolerance for the ambiguity that experienced humans systematically absorb. And as AI systems begin operating across functions, small differences in interpretation start to compound. If one part of the organisation tolerates a five percent error rate and another treats one percent as unacceptable, whose standard is the model running on?
The friction that follows is rarely about the intelligence of the model. It is about misaligned assumptions contained by proximity and pace. AI connects those contexts at speed and makes their differences visible. Organisations are forced to align how they coordinate before they have fully agreed why. The behaviour changes because the system requires it. The cultural story catches up later, if at all.
This is the railway pattern playing out again.
Railways didn’t invent time. They revealed that local time could not scale. AI isn’t inventing organisational ambiguity. It is revealing how much of organisational life depends on shared but unexamined interpretation and forcing that interpretation into the open.
But there is something else happening underneath that.
Look at the pattern across technology transitions. Railways standardised time. Industrialisation standardised production. Computing standardised logic. The internet standardised connectivity. Each transition forced something previously implicit into an explicit shared standard and each time, that standardisation created a surface others could build on. The rails didn’t just move goods, they made mass manufacturing viable.
If that pattern holds, perhaps the more interesting question about AI is not what it automates, but what it standardises.
The strongest candidate is interpretation itself. Pattern recognition. The modelling of uncertainty. The normalisation of probabilistic reasoning in everyday decisions. If AI embeds those things into organisational infrastructure; into thresholds, escalation paths, definitions of done and good enough. It is not simply a tool sitting on top of existing systems. It is laying track.
The people laying railway track were not consciously constructing the foundations of globally synchronised time. They were trying to move goods more efficiently. The wider consequences only became visible in hindsight, long after the track was fixed.
Railway time wasn’t just an operational fix. It was a new relationship between humans and time itself. Within a generation, nobody remembered negotiating it.
The standards being written into AI systems now will not feel like infrastructure. They will feel like sensible defaults. But every default encodes a judgement about what normal looks like and whose interests it serves. Within a generation, nobody will remember negotiating them. The question is whether we are paying attention while there is still something to negotiate.