Alba Lavagnoli Gonzalez
Universidad de Valladolid
Gilbert Ryle (1949) introduced a fundamental distinction between “knowing that” and “knowing how”, defining the latter as a type of practical or dispositional knowledge. Starting from this dichotomy, this paper aims to investigate whether the knowledge generated by machine learning and deep learning can be classified precisely as “knowing how”. Although authors such as Donald Gillies (2024) have already raised this question without reaching a definitive conclusion, we aim to determine whether these systems produce genuine practical knowledge or are limited to the execution of mere rote skills. In this analysis, we will draw on the contribution of Cameron Buckner (2023), who has highlighted how deep learning architectures exhibit profound analogies with human cognitive processes. A focal point of the research will be the analysis of the “opaque” nature of such systems: it is hypothesized that the “black box” nature typical of deep learning reflects the rylean idea according to which knowing how to do things does not necessarily require the ability to explain or justify the logical rules underlying the action.
Chair: Tyler Miniati
Time: 12:00 – 12:30
Location: SR 1.006
SOPhiA 2026 