Mikael Brunila: AI and the Struggle over the New Basic Units of Digital Life

“More transparency!”, reads a common demand in current debates on how to democratize AI as it spreads into our everyday lives. At Käänne, Mikael Brunila will, by contrast, propose that the real stakes in AI can only be addressed by creating the means for ordinary people to use AI locally, while hiding from the centralized gaze of large corporations and governments: An opacity from below.

Today, large language models (LLMs) have been adapted to abstract our everyday lives in a wide-range of domains, from representations of text and images to behavioral profiles of swipes on dating apps, listening preferences on streaming services, clicks in browser sessions, and DNA-sequencing. Through a historical contrast with the Pirate movement in Sweden between the fall of Napster and the rise of Spotify, Brunila suggests that the real issue with LLMs isn’t that they abstract, but the centralization of these powers of abstraction.

Instead of “one model to rule them all”, we need a million models ruled by all. LLMs should be developed locally and protected by similar cryptography that has become standard in applications like instant messaging. Indeed, it is only through claiming the AIs of our lives that we can avoid a dystopian future where the proliferation and development of AI becomes synonymous with ubiquitous and centralized surveillance.

Mikael Brunila is a postdoctoral researcher in geography at McGill University, specialising in philosophy of science and quantitative methods. His PhD thesis explored the critical application of information science in the social sciences, both in the field of language models and urban policy. Brunila has published four books, most recently ”Kuulumme toisillemme” (Khaos, 2023). In the 2000s, he was active in the squatter movement in Helsinki.

Photo: Henri Salonen