Loading...
Description
Artificial Intelligence, i.e., complex algorithms that learn to perform functions associated with human minds, such as perceiving, decision-making, and demonstrating creativity. Indeed, more often than not, AI is trained on biased datasets, that is, this data is disproportionately weighted in favor of or against certain individuals or groups of individuals. The roots for such biases are very diverse, sometimes they are technical in nature, but often they originate in the minds of people, making it difficult to identify these biases, e.g. if such a disproportionate weighting is perceived as ‘normal’ by many, while it is still devastating to few. We argue that the use of virtue ethics in AI can help to mitigate the consequences of biases and to help identify such biases. In particular, we aim to assess in multiple online and field experiments how the virtue ‘transparency’ affects an individual’s decision-making and perceptions regarding the AI.
Recommended Citation
Frenzel, Adeline; Jain, Shilpi; Jia, Shizhen (Jasper); Welck, Maximilian; and Langer, Nishtha, "Fighting the real AI Danger: How to Design Virtuous AI for Virtuous Decision-making" (2020). ICIS 2020 Proceedings. 2.
https://aisel.aisnet.org/icis2020/paperathon/paperathon/2
Fighting the real AI Danger: How to Design Virtuous AI for Virtuous Decision-making
Artificial Intelligence, i.e., complex algorithms that learn to perform functions associated with human minds, such as perceiving, decision-making, and demonstrating creativity. Indeed, more often than not, AI is trained on biased datasets, that is, this data is disproportionately weighted in favor of or against certain individuals or groups of individuals. The roots for such biases are very diverse, sometimes they are technical in nature, but often they originate in the minds of people, making it difficult to identify these biases, e.g. if such a disproportionate weighting is perceived as ‘normal’ by many, while it is still devastating to few. We argue that the use of virtue ethics in AI can help to mitigate the consequences of biases and to help identify such biases. In particular, we aim to assess in multiple online and field experiments how the virtue ‘transparency’ affects an individual’s decision-making and perceptions regarding the AI.
When commenting on articles, please be friendly, welcoming, respectful and abide by the AIS eLibrary Discussion Thread Code of Conduct posted here.