Between 1964 and 1968, Joseph Weizenbaum created ELIZA in the MAD-SLIP language, a pattern-matching program that simulated conversation. Its logic was declarative, legible, readable, and most interestingly, scripted. The source code was an important part of understanding how it worked. But then the source code was lost for 60 years and only a description of the algorithm remained, until I and a team of researchers rediscovered it in the MIT archives.

This talk takes the recently excavated ELIZA source code as a point of departure for understanding a transformation in what computation is and how it works. Contemporary AI systems do not operate by a defined script. They convert language, image, and sound into positions within high-dimensional geometric spaces. Qualitative difference becomes quantitative distance. Meaning has not disappeared but it has been replaced by geometric position within a proprietary manifold that cannot be read, contested, or audited the way a program might be. Drawing on critical code studies and critical theory, this talk explores the shift from script to vector and argues that we are witnessing a transition away from the centrality of the digital medium to a vector medium, with consequences for how language, knowledge, and media are produced, distributed, and consumed.

Bio

David M. Berry is Professor of Digital Humanities at the University of Sussex and Otto Mønsted Visiting Professor at Copenhagen Business School. His books include The Philosophy of Software (Palgrave, 2011), Critical Theory and the Digital (Bloomsbury, 2014), the co-written Inventing ELIZA: How the First Chatbot Shaped the Future of AI (MIT Press, 2026), and the forthcoming Artificial Intelligence and Critical Theory (Manchester University Press, 2026). His current research develops a critical theory of the vectorial turn, examining how high-dimensional geometric computation creates a new vector medium which transforms meaning, knowledge, and media.