Learned Synthetic Rationality Entity (LSRE): A Conceptual Proposal for the Classification of Non-Biological Rationality

Abstract

This article proposes the concept of Learned Synthetic Rationality Entity (LSRE) as a classificatory category for non-biological entities whose functional rationality is constituted through artificial processes of learning and training, becomes relatively stabilized in its inferential dispositions, and is instantiated in an artificial system. The proposal distinguishes rationality from phenomenal consciousness, arguing that the attribution of functional rationality does not, by itself, entail consciousness, subjective experience, human equivalence, or moral personhood. The article develops six dimensions of the LSRE: synthetic constitution, functional rationality, acquisition through learning, functional stabilization, instantiated entity, and independence from phenomenal consciousness. The article also proposes epistemic criteria for the attribution of rationality, including inference, coherence, consistency, evidential adequacy, generalization, and revisability. Particular attention is given to the distinction between historical dependence upon human-generated data and practices and identity with any particular human thought. The contribution is conceptual and classificatory. It seeks to provide a precise vocabulary for discussing forms of non-biological, learned, and functionally manifested rationality without prematurely resolving questions concerning artificial consciousness, subjectivity, agency, or moral status.