What makes the trajectory structural rather than contingent is that it follows from the substitution principle combined with the incentive structure of competitive markets. At each stage, the enterprise has an incentive to invest in further automation of the human functions that remain, because those functions are the largest remaining cost in the productive process. Each investment thins the human role further. The thinning is not a policy choice; it is the cumulative effect of many small investments, each locally rational, whose aggregate trajectory is the degradation the original workers did not want.
The trajectory is experientially deceptive because each stage feels liberating in the moment. The minder who no longer has to diagnose thread breakages is genuinely relieved to be free of a tedious task. The developer who no longer has to write boilerplate code is genuinely liberated to work on more interesting problems. The relief and the liberation are real; they are not errors of perception. But they are also the experiential surface of a structural transformation whose cumulative effect — over years or decades — is the progressive reduction of the human role to a thinness that previous generations would have found insulting.
The AI transition's timeline is the distinguishing feature. The power loom's trajectory from skilled partnership to mere overlooking took roughly eighty years. The assembly line's took sixty. The desktop publishing trajectory took twenty. AI tools are improving on a scale measured in months, and the compression of the trajectory may produce the transition from stage one to stage three within a single decade. The compression matters because institutional adaptation — retraining, social safety nets, educational reform — operates on timelines measured in decades. The mismatch between the speed of displacement and the speed of institutional response is the central policy problem of the AI age.
You On AI's implicit hope is that the trajectory can be arrested at stage one — the skilled partnership stage — through institutional design. The historical record is not encouraging. No previous industrial trajectory has been successfully arrested at an intermediate stage. The institutions that eventually ameliorated the worst consequences of each transition did so by providing safety nets for the displaced, not by preventing the displacement. Whether AI's compressed timeline permits a different pattern is the central empirical question of the next decade.
The trajectory is implicit in Ure's substitution principle but was not stated by him as a three-stage arc. The three-stage formulation is a reconstruction of the empirical pattern observed across industries since 1835, visible with particular clarity in the textile industry Ure studied, the automotive industry Ford pioneered, and the printing industry that desktop publishing transformed.
Stage one: skilled partnership. The machine is new, limited, and dependent on human expertise to function effectively; the partnership is genuine and the human contribution is substantive.
Stage two: routine monitoring. The machine has improved; the human's role has thinned from active practice to passive observation, with intervention required only for anomalies.
Stage three: mere overlooking. The machine operates autonomously; the human is retained for liability and residual anxiety rather than for positive contribution.
The compression effect. Each subsequent industrial trajectory has taken less time than the previous one; AI's trajectory may compress the full arc into a single decade.
The institutional mismatch. Adaptation — retraining, safety nets, education reform — operates on timescales that cannot match the compressed trajectory, producing a widening gap between displacement and response.