Machine Unlearning Doesn't Do What You Think: Lessons for Generative AI Policy and Research
#1307 · A. Feder Cooper, Christopher Choquette-Choo, Miranda Bogen, Kevin Klyman, Matthew Jagielski, Katja Filippova, Ken Liu, Alex Chouldechova, Jamie Hayes, Yangsibo Huang, Eleni Triantafillou, Peter Kairouz, Nicole Mitchell, Niloofar Mireshghallah, Abigail Jacobs, James Grimmelmann, Vitaly Shmatikov, Christopher De Sa, I Shumailov, Andreas Terzis, Solon Barocas, Jennifer Wortman Vaughan, danah boyd, Yejin Choi, Sanmi Koyejo, Fernando Delgado, Percy Liang, Daniel Ho, Pamela Samuelson, Miles Brundage, David Bau, Seth Neel, Hanna Wallach, Amy Cyphert, Mark Lemley, Nicolas Papernot, Katherine Lee
We argue that there are important gaps between what unlearning methods can do and claims about/hopes for what these methods can do for broader impact.
