The Financial Times has covered our recent report Toward Trustworthy AI: Mechanisms for Supporting Verifiable Claims. It quotes Haydn Belfield:
The new report reflects a shift from more abstract ethical concerns towards actionable solutions. “With all these ethics principles, we wanted to know how to tell if these are really true and if [developers] were really keeping to them,” said Haydn Belfield, project manager at Cambridge university’s Centre for the Study of Existential Risk. “That’s the real problem we set out to solve.”
According to Mr Belfield, incentivising users to carefully scrutinise AI systems could help catch problems earlier in the development process, although the report notes that this does not absolve developers of responsibility.
Toward Trustworthy AI: Mechanisms for Supporting Verifiable Claims
Report by Miles Brundage, Shahar Avin, Jasmine Wang, Haydn Belfield, Gretchen Krueger, Gillian Hadfield, Heidy Khlaaf, Jingying Yang, Helen Toner, Ruth Fong, Tegan Maharaj, Pang Wei Koh, Sara Hooker, Jade Leung, Andrew Trask, Emma Bluemke, Jon Lebensold, Cullen O'Keefe, Mark Koren, Théo Ryffel, JB Rubinovitz, Tamay Besiroglu, Federica Carugati, Jack Clark, Peter Eckersley, Sarah de Haas, Martiza Johnson, Ben Laurie, Alex Ingerman, Igor Krawczuk, Amanda Askell, Rosario Cammarota, Andrew Lohn, Shagun Sodhani, Charlotte Stix, Peter Henderson, Logan Graham, Carina Prunkl, Bianca Martin, Elizabeth Seger, Noa Zilberman, Seán Ó hÉigeartaigh, Frens Kroeger, Girish Sastry, Rebecca Kagan, Adrian Weller, Brian Tse, Beth Barnes, Allan Dafoe, Paul Scharre, Martijn Rasser, David Kreuger, Carrick Flynn, Ariel Herbert-Voss, Thomas Krendl Gilbert, Lisa Dyer, Saif Khan, Markus Anderljung, Yoshua Bengio