This week I gave a talk as part of the Chaos Computer Club event FireShonks, and it is now online: “Chat Control: Mass Screenings, Massive Dangers” (Dec. 27, 2023). It warns that mass screenings for low-prevalence problems are often doomed to backfire according to the implications of probability theory. It builds on numerous recent SubStacks, including:
– this book review of Welch et al’s Overdiagnosed, which focuses on the dangers of medical programs that share this dangerous structure;
– this effort to situate possible system-level solutions in the context of existing legal regimes, especially the concept of suitability in EU/German law;
– these Chat Control posts (1, 2, 3, 4);
– these AI Act posts (1, 2, 3); and
– this essay on reification and the common liberty-security mismappings in relevant debates.
Happy new year to digital freedom fighters and techno-solutionists all!
