Dead Zone of Accountability: Why Social Claims in Machine Learning Research Should Be Articulated and Defended
Paper argues that unaccountable "social claims" in ML research create a gap between promised and actual social impacts, and proposes accountability lenses and research agendas.