Title: Proceedings of the 2021 AAAI/ACM Conference on AI, Ethics, and Society
Abstract: We are very pleased to welcome you to the 2021 AAAI / ACM Conference on Artificial Intelligence, Ethics, and Society - AIES 2021. This is the fourth edition of this conference, and it is being held virtually on 19 - 21 May 2021, because of the Covid-19 pandemic. Over the last few years, the world has awoken to the power that we have vested-often without thought or care-in the people and systems that collect, aggregate, analyze, and act on our data. At the same time, AI systems promise new ways to empower individuals and collectives to change society from the bottom up. International organizations, governments, universities, corporations, and philanthropists have recognized the urgent need to bring all of our intellectual tools to bear on charting a course through this uncertain new territory. Earlier iterations of this conference and others have seen the first fruits of these calls to action, as programs for research have been set out in many fields relevant to AI, Ethics, and Society. The early days of shaking us awake are done: we now know, well, that we are increasingly reliant on AI systems that are radically changing the world around us, for better and worse. The next step is to chart a course forward, both by deepening our diagnosis of where we are now, and by developing new goals, models, and technical and regulatory systems to shape the future of AI and society toward how we collectively intend our societies to look. The AIES-2021 Call for Papers attracted submissions from Asia, Canada, Australia, Europe, Africa, and the United States. The Main Track of AIES-2021 received 278 submissions, of which 106 (38%) were accepted, with full papers included the Proceedings and posters in the poster sessions for discussion. A subset of 27 of the accepted papers (9.7% of submissions) were selected for oral presentation as well. Plus, we have eight talented participants in our Student Track. Three keynote speakers are scheduled. Ifeoma Ajunwa, Associate Professor at the University of North Carolina School of Law, will speak on "A Veil of Ignorance for Automated Decision- Making". Timnit Gebru, co-founder of Black in AI, will speak on "Moving beyond the Fairness Rhetoric in AI." Arvind Narayanan, Associate Professor of Computer Science at Princeton University, will speak on "The Ethics of Datasets: Moving Forward Requires Stepping Back". An interesting aspect of running an interdisciplinary conference on AI, Ethics & Society is that ethical issues come up in the design of the conference. We certainly can't claim to have found good answers to every ethical issue, but we have tried to address two of them. First, we felt that making the conference accessible to people with disabilities must be a priority. In a virtual environment, sight and hearing disabilities are more limiting than, say, mobility. We chose Hopin to provide our virtual environment because it did a better job of accommodating these disabilities than the alternatives. Second, our authors and attendees come from all over the world. In normal times, they have different distances to travel; in these days of virtual meetings, they attend the conference sessions while living in different time-zones. For AIES-2021, we scheduled the conference, not by utilitarian maximization of convenience for the greatest number, but by a somewhat more Rawlsian effort to improve the situation of those worst off. In both of these cases, majorities pay a small cost for significant improvements for those worst off. These are small issues in conference design, but they illustrate much larger ethical questions faced by our society (and addressed by our research). Our choices are not presented as solutions, but as starting points for discussion. Nonetheless, in these small ways, we have skin in the game.