CEA Panel on Reproducibility

Date:

Joint with Marie Connolly (UQAM, Moderator), Abel Brodeur (UOttawa, I4R), Joëlle Pineau (McGill University and Meta), Hugo Larochelle (Mila and Université de Montréal)

Panel at Canadian Economics Association meetings, Montréal, Canada

Location

8:30am-10am, UQAM, Pavillon DS, 320, rue Ste-Catherine est, room DS-R510

Abstract

Recent AI developments, particularly large language models (LLMs), came out in full force in the last couple of years, with an increasing use by researchers in all fields. How is this changing the way economists handle the reproducibility of their research? This panel will be chaired by Marie Connolly, Data Editor at the Canadian Journal of Economics, and feature economists Lars Vilhuber, the Data Editor of the American Economic Association, and Abel Brodeur, the founder and chair of the Institute for Replication, as well as computer scientists Joëlle Pineau and Hugo Larochelle, who are both leading scientists in the field of machine learning and deep learning. Our discussion will cover the following themes: how can research that uses AI tools comply with journals’ data and code availability policies? Could robots test the reproducibility of articles? What are the best practices in the field of computer science? What can economists learn from computer scientists, and vice versa?