Capability Without Scarcity

Across four decades of building, Sebastian Thrun repeated the same structural move in domain after domain: identify a capability that relieves suffering or enables flourishing, find the moment where machine learning can absorb that capability from human exemplars, and deliver it without the scarcity that has always rationed it. The driving skill of a trained human driver is scarce, bound to the attention span and reflexes of an individual body; Stanley learned it from data and poured it into any number of vehicles simultaneously. The explanatory skill of a great teacher is scarce, bound to the hours and presence of an individual person; a massive open online course copied it and delivered it to a hundred and sixty thousand learners at once. The diagnostic acuity of a board-certified dermatologist is scarce, bound to the trained eye of one person seeing one patient; a neural network absorbed that acuity from a large image dataset and deployed it on any phone