Honors Program


Document Type


Date of this Version



Shardelow, C. 2021. Avoiding the Basilisk: an Evaluation of Top-Down, Bottom-Up, and Hybrid Ethical Approaches to Artificial Intelligence. Undergraduate Honors Thesis. University of Nebraska-Lincoln.


Copyright Cole Shardelow 2021.


This thesis focuses on three specific approaches to implementing morality into artificial superintelligence (ASI) systems: top-down, bottom-up, and hybrid approaches. Each approach defines both the mechanical and moral functions an AI would attain if implemented. While research on machine ethics is already scarce, even less attention has been directed to which of these three prominent approaches would be most optimal in producing a moral ASI and avoiding a malevolent AI. Thus, this paper argues of the three machine ethics approaches, a hybrid model would best avoid the problems of superintelligent AI because it minimizes the problems of bottom-up and top-down approaches while maximizing their advantages. After detailing the importance of discussing morality in ASI’s and outlining some necessary conditions, the problems of machine ethics will be considered, and arguments for and against the three approaches will be responded to.

Included in

Philosophy Commons