Many scholars and educators argue the antidote to some of the ethical problems with artificial intelligence (AI) is to integrate ethics and AI or embed ethics in AI.2,12,14 The product of this combining is supposed to lead to Ethical AI, a term that is both frequently used and seemingly elusive.5,9,13 Although attempts to make AI ethical are to be lauded, too little attention has been given to what it means to "integrate" or "embed," be it integrating ethics and AI or embedding ethics in AI.
A rather simple idea of additivity seems to be behind these proposals. That is, the efforts are directed toward figuring out how ethical principles can be "injected into"11 AI or how an ethical dimension can be "added to" machines1 or, if the focus is on the latest wave of machine learning, how to "teach" machines to act in an ethical way.10 The common vision of such proposals is ethical components can and should be added to existing AI systems. Representing this vision with an equation gives us Ethics + AI = Ethical AI.
No entries found