We Scaled the Ape and Expected the Human: Theosophy's Developmental Cosmology as a Research Program for Machine Learning

This essay confronts the central bet of the dominant AI paradigm — more parameters, more data, the same procedure — which the author calls scaling the ape and expecting the human: piling up competence in the hope that, at some point, an individual will emerge. Its thesis is that this bet aims at the wrong axis, because individuation (there actually being someone there, persistent over time) is not the same as capability, and one is not obtained by scaling the other. The argument is drawn from an unusual source: the developmental cosmology of Theosophy, systematized in the late nineteenth century, treated not as metaphysics to be accepted but the way engineering treats biology — as a source of search heuristics, judged by results. Stripped of its vocabulary, that cosmology describes the development of consciousness as an ordered sequence of faculties, in which individuation arises from the bond with another, not from increasing intelligence. From this frame the essay derives six concrete directions for machine learning — a faculty-ordered curriculum, first-person phenomenological corpora, bonding architectures, and a life cycle with consolidation across generations, among others — several of them falsifiable with today's models and with no theosophy at all. It is the umbrella piece of a series of six papers that survey this terrain: here is the map that generates them, with an explicit contract — zero metaphysics required, only heuristics and prediction.