“Ethical Design: An Artificial Intelligence Imperative” by Tilak Agerwala
Systems using Artificial Intelligence (AI) are rapidly growing in capability, impact, and influence. Used by millions of people, they offer a significant opportunity to improve human well-being. At the same time, examples of AI misuse and failures abound. AI technology has been used to manipulate voters and to spread disinformation and create social unrest. Failures include machine bias in criminal sentencing, flawed voice-recognition technology leading to the deportation of thousands of students, fatal driver-less car crashes, and surgical robot incidents. These scenarios can violate privacy, threaten security, and even kill, raising ethical issues. The impact of failures can be mitigated using ethical design approaches. Recognizing this state of affairs, governments, professional societies, and institutions are establishing guidelines and regulations for the ethical design of AI systems. The presentation describes the current AI landscape, provides an overview of Ethical Design, includes practical suggestions for AI developers (“everyday ethics”), and identifies open research questions in the areas of “ethically aligned AI” and “fully ethical agents.” In conclusion: AI designers and developers must build lawful, ethical, and robust systems, by practicing “everyday ethics”; government, academia, and industry must develop cross-disciplinary research programs in ethical design; and consumers of AI technology must raise their awareness of the potential for the misuse of AI.
Abstract Excerpt from “The Svasthya-Rasa-Bodhini” CSP public lecture on 29 November 2019, by Tilak Agerwala (Dr. S. Radhakrishnan Chair visiting professor at NIAS, Bangalore, and IBM Emeritus, New York.)
Disclaimer: The opinions endorsed by the speaker is solely his and not in any way endorsed by the organisers.