Workshop on Automated Decision-Making
Introduction
Decision making consists in choosing a solution from a set of alternatives. Decision making is therefore at the heart of the exercise of morality, since it directly precedes action. Today, AI can automate all or part of this process. This is referred to as fully-automated or semi-automated decision. The advantages of using this type of technology are obvious: the decision is fast, and it can be deployed at very large scales. But this automation raises many ethical and technical considerations, in particular concerning the dehumanization of this process as well as its reliability. Nevertheless, these tools are spreading fast to an increasing number of applications (education, medicine, marketing, bank loans, etc.) and has significant consequences for the people concerned.
This meeting confronted theory to practise, through presentations that addressed decision-making from different perspectives (computer science, medicine, law, cognitive sciences) and were be followed by interdisciplinary exchanges.
Program
Meeting of the “Ethics and Artificial Intelligence” group, organized on Wednesday May 12, 2021:
- Anita Burgun, AP-HP, Université de Paris, “Medical decision: concepts and realities”
- Jérôme Lang, Dauphine, PSL, “Collective decision: from ethics to axioms”
- Thierry Kirat, Dauphine, PSL, “Predictive justice or legal analytics? Predictive models and quantified case law, a variety of applications”
- Virginie Do, Dauphine, PSL, “Equity in Machine Learning”
- Sophie Bavard, ENS, PSL, “Decision making and reinforcement learning in humans”
The meeting was held online, on Zoom.