Series: PostScriptUM #56

In 2024, four MIT researchers named a hypothesis after Plato’s cave: neural networks, trained on different data and tasks, converge on one representation of reality. In 𝘊𝘢𝘷𝘦 𝘔𝘢𝘤𝘩𝘪𝘯𝘦𝘴, M. Beatrice Fazi takes the conjecture at its word, following it back to Plato, where it comes apart. The models never turn from the wall; the reality behind their agreement falls away. What remains is stranger than realism: representation as construction, structure built, not found, and a cave that is all there is.

𝗠. 𝗕𝗲𝗮𝘁𝗿𝗶𝗰𝗲 𝗙𝗮𝘇𝗶 is a philosopher working on the philosophies of computation, technology and media. She has published extensively on the limits and potentialities of the computational method, on digital aesthetics and on the automation of thought. She is Associate Professor at the University of Sussex (United Kingdom), author of 𝘊𝘰𝘯𝘵𝘪𝘯𝘨𝘦𝘯𝘵 𝘊𝘰𝘮𝘱𝘶𝘵𝘢𝘵𝘪𝘰𝘯: 𝘈𝘣𝘴𝘵𝘳𝘢𝘤𝘵𝘪𝘰𝘯, 𝘌𝘹𝘱𝘦𝘳𝘪𝘦𝘯𝘤𝘦, 𝘢𝘯𝘥 𝘐𝘯𝘥𝘦𝘵𝘦𝘳𝘮𝘪𝘯𝘢𝘤𝘺 𝘪𝘯 𝘊𝘰𝘮𝘱𝘶𝘵𝘢𝘵𝘪𝘰𝘯𝘢𝘭 𝘈𝘦𝘴𝘵𝘩𝘦𝘵𝘪𝘤𝘴 (2018) and co-author of 𝘋𝘪𝘨𝘪𝘵𝘢𝘭 𝘛𝘩𝘦𝘰𝘳𝘺 (2025).