Causal Data Science Meeting 2022: Using Causal Inference to Solve Problems We Care About
Posted November 24, 2022 by Paul Hünermund and Jermain Kaminski and Beyers Louw and Carla Schmitt ‐ 4 min read
We are delighted to share a summary of the Causal Data Science Meeting 2022 with you. The event was jointly organized by Maastricht University and Copenhagen Business School.
About the Meeting
On November 7-8, Maastricht University and Copenhagen Business School attracted more than 1,900 virtual attendees and welcomed the 2011 Turing Award Winner Judea Pearl as well as fairness and causality expert Silvia Chiappa, Group Leader of Causal Intelligence team at Google DeepMind. Now in its third year, the Causal Data Science Meeting 2022 #CDSM22, featured a wonderful lineup of speakers, two keynotes and an industry panel. The conference was established to bring together researchers and experts from academia and industry to engage in discussion and push the boundaries of causal inference. This year saw an increase in both submissions and registrations from the previous two iterations. The lineup included 32 leading researchers from academia and data scientists from industry. The conference attracted about 60% academic, and 40% industry attendees, representing most US companies at the forefront of data science.
Applications
We saw valuable applications of causal inference in fields such as medicine, policy, strategy and more. Although we have seen progress since the establishment of the conference, it is still clear that we face a challenging albeit exciting road ahead as Ian McCormick - who presented an application of causal diagrams in the early detection and diagnoses of Alzheimer’s – early noted in first round of presentations: “A causal revolution is needed in medicine”. The managerial challenges that come with solidifying organizational expertise into causal models is a theme that was felt throughout the conference.
“The next revolution will be even more impactful upon realizing that data science is the science of interpreting reality, not of summarizing data.” – Judea Pearl, Recipient of the Turing Award in 2011 and Author from The Book of Why, UCLA
Keynotes
Research presentations were complemented by two excellent Keynotes. For one, Professor Judea Pearl from UCLA (Youtube). In 2011, the Association for Computing Machinery awarded Pearl with the Turing Award, the highest distinction in computer science for fundamental contributions to artificial intelligence through probabilistic and causal reasoning. The conference was also honored to welcome Professor Silvia Chiappa, research scientist and group leader of the Causal Intelligence team at DeepMind London, and Honorary Professor at the Computer Science Department of University College London. A motivating insight was shared from Silvia Chiappa when asked what the “Killer Application” of Causal Inference will be, she noted that in industry and practice there are many problems that they work on at DeepMind and each solution has its contribution to the whole. Implying that the groundwork in causal inference is important and each insight that we put forth in the field has an impact.
State of Industry
A panel discussion of industry leaders completed the picture, moderated by Victor Chen (Fidelity Investments) who was joined by industry leaders of data science and experimentation including Sathya Anand (Netflix), Somit Gupta (Microsoft), Mikael Konutgan (Meta), YinYin Yu (LinkedIn), Benjamin Skrainka (eBay) and Eric Weber (Stitch Fix). Somit Gupta from Microsoft described that the importance of causal inference manifests on two levels. The first level is to validate experimentation results so that they can be more confident and limit the risk in their decision-making capabilities “We need to be confident that our decision causes X, Y, Z and doesn’t harm our users”. The second level is about the cultural impact of applying causal knowledge at scale “This allows us to make decisions that are more consistent at the organizational-level and makes our decisions more scrutinizable as our key-metrics are known”. All-in-all, describing a bright and exciting future for the importance of causal inference tools. You can view a record of the panel discussion right here (YouTube).
We would like to give a heartful thank you to our sponsor Geminos. We thank the presenters and panelists for presenting inspiring ideas and all the discussants for thought provoking questions.
What’s Next?
The Causal Data Science Meeting 2023 is scheduled for two days in November 2023 (very likely November 6–7, or 7–8). All registred participants for the 2022 conference will be updated accordingly. We further consider running a smaller in-person roundtable with industry in the near future. If you like to participate, feel free to reach out to contact@causalscience.org.
Header image by Aakash Dhage on Unsplash