MoralNet: A Benchmarking Protocol for Evaluating Morality in AI Systems

Aníbal Monasterio Astobiza · 2023

Artificial intelligence has prompted unprecedented concerns regarding ethical decision-making in autonomous systems. Here, we introduce MoralNet, a proof ofconcept database and benchmarking platform akin to ImageNet, designed to assess, evaluate, and iteratively refine any instance of ethical decision-making in machines.Inspired by the lessons of moral psychology and moral philosophy, MoralNet is a call to action to take the first step towards a computational ethics.

Read the paper · More papers on PaperTik