Filling gaps in trustworthy development of AI

Science. 2021 Dec 10;374(6573):1327-1329. doi: 10.1126/science.abi7176. Epub 2021 Dec 9.

Shahar Avin¹, Haydn Belfield^{1

2}, Miles Brundage³, Gretchen Krueger³, Jasmine Wang⁴, Adrian Weller^{2

5

6}, Markus Anderljung⁷, Igor Krawczuk⁸, David Krueger^{5

9}, Jonathan Lebensold^{4

9}, Tegan Maharaj^{9

10}, Noa Zilberman¹¹

¹ Centre for the Study of Existential Risk, University of Cambridge, Cambridge, UK.
² Leverhulme Centre for the Future of Intelligence, University of Cambridge, Cambridge, UK.
³ OpenAI, San Francisco, CA, USA.
⁴ School of Computer Science, McGill University, Montreal, QC, Canada.
⁵ Department of Engineering, University of Cambridge, Cambridge, UK.
⁶ The Alan Turing Institute, London, UK.
⁷ Centre for the Governance of AI, Oxford, UK.
⁸ École Polytechnique Fédérale de Lausanne, Lausanne, Switzerland.
⁹ Mila, Montreal, QC, Canada.
¹⁰ Faculty of Information, University of Toronto, Toronto, ON, Canada.
¹¹ Department of Engineering Science, University of Oxford, Oxford, UK.

Abstract

Incident sharing, auditing, and other concrete mechanisms could help verify the trustworthiness of actors.