An autonomous vehicle must decide whether to veer left or right. If it goes left, it hits an elderly woman, and if it goes right it hits a young boy. Which way should the car go? Why? These are the kinds of dilemmas we will explore across domains from AI healthcare, to hiring tools, to the recommendation algorithms of social media, and across the range of today’s AI applications.
To respond to the dilemmas, the core principles employed in today’s AI ethics will be developed. They are: Autonomy, Human Dignity, Privacy, Fairness, Equity, Social Wellbeing, Explainability, Safety, Performance.
Teaching methods
The teaching method is classroom discussion of case studies, supplemented by lectures from the professor. There are no required texts and no homework - but attendance at seminar sessions is required because the course's main ideas will be developed collaboratively, through the seminar discussions. AI ethics will be learned by doing AI ethics.
Assessment methods
Students will present a power point / poster presentation. It will be an AI ethics evaluation of an AI application. The AI application may be a tool the student is developing in their own work, or it may be a publicly known artificial intelligence application (ChatGPT, for example, or smart glasses, or Tesla and autopilot). The presentation will last 15 - 20 minutes plus 5 - 10 minutes of questions.
Students will be graded on their ability to locate the ethical dilemmas that arise around AI technology, and their ability to discuss the dilemmas knowledgeably. There are no right or wrong answers in ethics, but there are better and worse understandings of the human values that guide and justify decisions.
Because the main ideas will be developed through classroom discussion, attendance to at least 80% of seminar sessions is required in order to do the final presentation.
Bibliography
The bibliography will be the seminar sessions and the subsequently published decks.
Website: https://trento.ai.ethicsworkshop.org/
