Convivial Machine Learning (DHSI 2026)

Description

Ivan Illich wrote of the alphabet and the printing press that they are “almost ideally convivial” because “anybody can learn to use them, and for [their] own purpose. They use cheap materials. People can take them or leave them as they wish. They are not easily controlled by third parties” (Illich, Tools for Conviviality, 1973). Yet, these affordances of the alphabet and the printing press, as described by Illich, came after centuries of the parallel development of technical innovations and social practices which made these technologies convivial.

In this course, we will take a historical and speculative route to interrogate what lessons from manuscript and print culture we can apply to design and think machine learning tools that are open and convivial. We can draw a straight line from the alphabet (the discretization of speech sounds into letters) and the printing press (the first mechanical means of textual (re)production) to machine learning. Through a series of digital and analog experimental projects, we will situate machine learning within a history of radical and innovative writing technologies that can serve as models for designing and thinking more convivial machine learning systems.

Specifically, students will learn how to use open-source language models locally, build their own small-scale language models, and learn how to use open-source and powerful transcription models. They will also learn how to document their process digitally and materially with small, hand-made publications. Through this mix of historical inquiry and hands-on experimentation, students will have developed practical skills in working with open machine learning tools and developed a reflection on how to design interactions and technologies that enhance human autonomy and creativity.

No prerequisites are required for this class.

Instructor(s)

Gabrielle Benabdallah is a Sloan postdoctoral fellow at the University of Washington in Seattle, where her research explores the materiality of knowledge production, from print culture to artificial intelligence. With a background in comparative literature and textual studies, she has dedicated the past seven years to working and publishing in the fields of human-computer interaction and interaction design. Currently, she is focused on examining the influence of AI-augmented tools on practices in scientific publishing.