Considerations for ethical AI include clarity or accountability regarding the purposes of technology, and its role in decision processes. Clarity underpins responsible use of technology, and informed engagement (and consent) with it. Accountability of AI's use by humans, to humans underpins respect for persons. However, achieving clarity and accountability may be challenging where there are different levels of expertise regarding the technology, and understandings of the contexts into which technology may be deployed. An aim of learning environments should be to develop agency, but this may be stymied by systems that - however accurate - may not adequately explain decisions so they are understood to a wide range of learners, or that limit choices for either teachers or learners.Areas of technology development and use, as other areas of applied research, inherently involve understanding both of the technical features of research (methods, tools, etc.), and its practical context including stakeholders. This kind of research requires engagement of expertise from stakeholders in the contexts into which AI will be deployed, and of research expertise from both technical and social disciplines. This involvement should not be siloed, instead there should be engagement across those involved. This can be challenging given the different areas and levels of expertise each group may have. Understanding of the ethical issues arising from research is one area where this transdisciplinary work is important. Effective development of AI for use in society requires a level of 'AI literacy' among stakeholders, engagement with transparency and explanation among providers, and an understanding of the - dynamic - social context into which any technology may be deployed. "More institutional resources and incentive structures are necessary to bring A/IS engineers and designers into sustained and constructive contact with ethicists, legal scholars, and social scientists, both in academia and industry. This contact is necessary as it can enable meaningful interdisciplinary collaboration and shape the future of technological innovation. More could be done to develop methods, shared knowledge, and lexicons that would facilitatesuch collaboration.This issue relates, among other things, to funding models as well as the lack of diversity of backgrounds and perspectives in A/IS-related institutions and companies, which limit cross-pollination between disciplines. To help bridge this gap, additional translation work and resource sharing, including websites and Massive Open Online Courses (MOOCs), need to happen among technologists and other relevant experts, e.g., in medicine, architecture, law, philosophy, psychology, and cognitive science. Furthermore, there is a need for more cross-disciplinary conversation and multi-disciplinary research, as is being done, for instance, at the annual ACM Fairness, Accountability, and Transparency (FAT*) conference or the work done by the Canadian Institute For Advanced Research (CIFAR), whichis developing Canada’s AI strategy.Funding models and institutional incentive structures should be reviewed and revised to prioritize projects with interdisciplinary ethics components to encourage integration of ethics into projects at all levels.
##
# Further Resources- S. Barocas, Course Material for Ethics and Policy in Data Science, Cornell University,
1. - L. Floridi, and M. Taddeo. “What Is Data Ethics?” Philosophical Transactions of the Royal Society, _vol. 374, no. 2083, 1–
1. DOI
1. 1098/ rsta.
1. 0360,
1. - S. Spiekermann, Ethical IT Innovation: A ValueBased System Design Approach. Boca Raton, FL: Auerbach Publications,
1. - K. Crawford, “Artificial Intelligence’s White Guy Problem”, _New York Times, July 25,
1. [Online]. Available: http://www.nytimes. com/2016/06/26/opinion/sunday/artificialintelligences-white-guy-problem.html?_r=1. [Accessed October 28, 2018]."p.123-124