Explainable AI – Transparency and Accountability in a Data-Driven Society
Tuesday April 9th, 9.30 - 19.00
The impressing increase in data-driven services and tools that use Artificial Intelligence, Machine Learning, Data Mining and Data Visualization, has given rise to concerns on the accuracy, validity and transparency of the underlying models, assumptions and data sources. Often referred to as Explainable AI, recent research discusses issues related to transparency and accountability of such “smart” systems. This workshop brings together experts from different fields that study how to provide transparency, explanations and accountability to models, services and systems that create a representation of human behaviour and actions, and that increasingly impact our daily life.
The workshop will present:
(1) an overview of criticism on existing systems or models,
(2) approaches to provide explanations and discrimination-aware analysis methods, and
(3) user experience, interaction and design approaches for smart interactive systems and Big Data approaches in general.
09.30-09.45 – Opening: Introduction of the workshop and overall objectives
The Autonomous Internet of Things and Explainable AI by Enrico Costanza, University College London
Designing Human-Centric Explainable AI by Brian Y. Lim, National University of Singapore
11.15–11.30: Coffee break
Explainability through Abductive Hypothesizing: Understanding and improving our models through open-ended investigation by Simon Enni, Aarhus University
Panel with Enrico, Brian, and Simon
13.00–14.00: Lunch in the Incuba Katrinebjerg Canteen
Explainable Machine Learning is Often More Complex and Less Helpful Than You Might Think: Lessons from the Law, Lab and the Office by Michael Veale, University College London
Making Conclusions Conclusive: Challenges of Interpreting Machine Learning Results for Science by Indrė Žliobaitė, University of Helsinki
15.30-15.45: Coffee break
Machine-Assisted Decision-Making: Understanding and Accounting for Human Factors by Nina Grgić-Hlača, Max Planck Institute for Software Systems
Panel with Michael, Indrė, Nina and Anne Henriksen
17.45-18.00: Coffee break
18.30–19.30: Closing Reception
Ira Assent, Professor in Data-Intensive Systems
Jo Vermeulen, Assistant Professor in Ubiquitous Computing and Interaction
Marianne Graves Petersen, Associate Professor in Ubiquitous Computing and Interaction
Audience / Target Group
The workshop is open to a general audience, and we welcome participation from researchers, students, and experts in industry working with data analysis.
Maximum 75 participants
Tuesday at 9.30 at Room Kahn 119K (5126-119K), Finlandsgade 26, Katrinebjerg, 8200 Aarhus N.