Applying AI ethics Fairness Ensuring to have the diverse database, no bias Accountability Clear responsibility defined for each AI system’s outcome When there is a malfunction, who is the responsible for that malfunction Transparency Make the AI system explainable and understandable Commitment Ethical adherence builds trust, and mitigates risks Privacy-personalization Personalization can compromise the privacy Transparency-complexity Easy model – easy to understand but less accuracy Complex model – hard to understand but improve the accuracy Autonomy-control Which one should be set up as the first priority? Control compromises autonomy