Sonia Bernac

Dr Sonia Bernac is a philosopher of media and science, artist, and creative technologist. Her research examines technologies of thought and systems of knowledge, with particular attention to the tensions between competing epistemic paradigms.

Her doctoral thesis at the Royal College of Art, A Bestiary of Distributed Intelligence, explored how inherited conceptions of computation shape and constrain the design of AI infrastructures.  It considers how earlier scientific schemas, abstractions, technical bottlenecks, common-sense assumptions, and residual cosmologies persist within contemporary architectures, evaluation procedures, and interface conventions. Her work attends to the often incommensurable frameworks that coexist within them and asks how their inherited assumptions might be unsettled or made available for critique and redesign.

Bernac examines the limits of both classical materialist traditions and the new materialisms in accounting for the strange materialities of digital systems. Drawing on complexity science, she considers how materialist thought might be revised in response. Her practice extends these questions through experimental AI systems that place different models in conversation with one another. She uses the resulting frictions, misrecognitions, and unexpected translations to investigate how heterogeneous computational logics interact.

She was previously a Senior Researcher at Antikythera and a researcher at the Artificial Intelligence in Design Laboratory (AiDLab), a joint initiative between the Royal College of Art and the Hong Kong Polytechnic University. She has delivered invited talks and exhibited her work internationally, and has advised research institutions on AI ethics, with particular attention to generative media. Recent talks and projects include Synthetic Disobedience: Misalignment and Incommensurability in Complex Artificial Systems (2026), Synthetic Mood, or the Aesthetic Question of Ethics (2026), and Deviations of a Fruit Fly (2024). She is also a registered expert with the European Commission.

Title of the talk: Against Synthetic Common Sense: The Computational Afterlives of Democratic Metaphors

Old technological worlds are not neatly disassembled before new ones are installed. Technological change proceeds through sedimentation, as new systems are retrofitted to legacy standards, interfaces, hardware, institutions and classificatory habits designed for earlier worlds. New technologies carry outdated technical and philosophical frameworks forward, embedding older diagrams of intelligence, order and collective life within emerging computational arrangements. Democratic imaginaries have accumulated particularly dense deposits of this conceptual detritus, combining images of the network and the political spectrum, models of the public sphere, figures of revolution and social fabric, and normative commitments to transparency, plurality and coherence. These figures and commitments emerge from distinct philosophical and technological histories and encode assumptions about what constitutes a system, how its parts should relate, how participation should be organised and why particular forms of connection acquire ethical value.

As these concepts enter discussions of AI, their historical specificity is compressed into moral shorthand, through which decentralisation comes to stand for democracy, transparency for accountability, plurality for inclusion and alignment for ethical order, while the philosophical work required to justify these associations is displaced by the apparent obviousness of the terms themselves. Using concrete examples from generative AI and planetary computational infrastructure, this talk traces these portable diagrams of thought across architectures, interfaces, datasets and evaluation procedures, examining how they organise the design of emerging systems as well as the vocabularies through which those systems are criticised.