Ethical AI Frameworks: A Bibliometric Study of Governance, Trust & Security

Abstract

Artificial Intelligence or AI is changing everything in our lives and how we live and work by the fact of being able to reason, interpret, decide, and solve problems that are impossible for humans to solve. But, as the technologies increasingly find their ways into spaces such as health care, finance, and government, concerns are being raised regarding the social, and particularly ethical, implications of these systems with regard to issues of fairness, accountability, and privacy. How do we build for AI to develop on an ethical, responsible, trustworthy, safe and where people will be comfortable with it foundation is addressed in this paper. Despite the promise, AI scares us, and with good reason. Those who compile this new kind of information are apt to, and often will, get things wrong or put out misleading or irrelevant information; there is ample opportunity for differential treatment and redress can be nearly impossible. As a result, “designing artificial techniques that people trust can use” is just as important as “designing them accurate or fast”. This implies that the ways in which they are produced has to be open, understandable and fair and that the data that the techniques process is well documented and safeguarded. The dissertation relies on cutting edge scholarship to explore the concrete practice of engineering responsible ai. It involves creating ways to minimize structural bias and make their work more transparent, as well as looking at resources like the Data set cited above, a means to extract the data. It is also a plea for public spaces where we, as designers, policy makers, and human beings more broadly, can collectively strive to bring A.I. into existence for the collective good. Rather, the technology can be made compatible with the human ethos and the liberal ethics system such that it instills in the AI the will to be of use to our species.

Ähnliche Arbeiten