Making Artificial Intelligence More Sustainable: Three Points of Entry into an Ethical Black Box

Résumé 0

The technological leaps in artificial intelligence (AI) over the past twenty years have profoundly renewed its ability to support, if not replace, humans in many settings, further intensifying rising concerns about decisions made, actions taken, and their potential consequences. A robust conceptual framework to engage with AI ethics is thus more necessary than ever in order to make it more sustainable. To this purpose, this essay advances an approach to AI ethics ‘from within’, defining it as the style of its algorithm(s) in practice. To demonstrate its practical value, I explore three points of entry: (1) value-laden patterns embedded in datasets used in machine learning, (2) the importance of value functions in the training and operating of AI, and (3) the possibility of adjusting some ‘ethics settings’. The example of algorithmic Human Resource Management (HRM) is examined to see how it can be brought closer to sustainable HRM. JEL Codes: O310