Assumptions of Computer Science Interacting with Assumptions of Text-driven Normativities
summary
Argues that privacy as protection of the incomputable self means formal tools like differential privacy and homomorphic encryption only solve reidentification and statistical leakage, not post‑inference harms; differential privacy lets builders claim an individual “doesn’t matter” to aggregates, which legitimizes large profiling systems and enables preemption via nudging (example: an 83% inferred chance of becoming vegetarian lowered to 43% after targeted messaging). Recommends agonistic machine learning: include domain experts and affected people early to expose trade‑offs.
AI-written summary · about
chapters (4)
related talks
01h 16mLaw and CS
27mThe ethical algorithm
33mPractical Data Privacy
01h 14mPerspective: Achieving Meaningful Privacy in Technology and Design
29mPrivacy-preserving methods: Building secure projects
05mPrivacy, identity, and autonomy in the age of big data and AI
01h 20mUnlocking societal benefits of AI with privacy protective technologies.
44mHybrid Legal-Technical Concepts of Privacy