SDSCon 2022 explores “the challenge of the era”: societal scale data systems
Over 120 attendees, academics and industry members gathered on April 1, 2022 for the fourth annual SDSCon, and the first SDSCon after the event was postponed by the pandemic in 2020 and 2021.
SDSCon 2022 featured a day of talks on a variety of topics, from machine learning and microeconomics to strategies for mitigating inherent biases in data and using statistics in astrophysics to explore the nature of dark matter.
Twenty MIT students from various departments presented posters at the event, highlighting applications in statistics and data science from areas such as information privacy, collective decision making, astronomy and neural networks. Adityanarayanan Radhakrishnan (EECS) won the Best Poster Award for his paper “Wide and Deep Neural Networks Achieve Optimality in Classification”
Watch talks and panels from SDSCon here or on Youtube.
“So the real question of AI to me is not, ‘how do you put thought in a computer?’ The real question is how do you build societal scale systems, planetary scale systems that are based on data? People’s preferences, multiple decisions, and all the tangles of all that. And they really work, they really respond well. They don’t give you nonsense. They don’t give you bad performance in the tails, et cetera. That really is the challenge of the era.” Michael Jordan, On Dynamics-Informed Blending of Machine Learning and Game Theory
“I hope that we can dig in a little deeper, and that you can think a little bit reflectively as you go back to your work, to the amazing research that you’re doing, and get a sense of: where is a place, where is a spot where I hadn’t really thought about? Where I might not be seeing the full picture, a deeper picture? And what is that saying? What is that doing to my conclusions, to my work?” Elisa Celis – Mitigation of Bias in Data Science
“I think that with the rate at which everything is changing around statistics and data science, our educational responsibilities are huge… Generally new technologies, particularly that those that are so broadly applicable, bring not only important advances, but also new challenges. Systems for making predictions and drawing conclusions from data change the operations of our businesses, the nature of our leisure activities and may go so far as to change our society and impact our image of our own humanness. The issues are all the more critical for systems that make autonomous decisions.” Dean Huttenlocher, Schwarzman College of Computing
