School of Computer Science Hosts Inaugural Seminar on Explainable AI
Researchers, academics and postgraduate students gathered for the inaugural Seminar on XAI, hosted by the TU Dublin School of Computer Science
The Centre of eXplainable Artificial Intelligence, together with the School of Computer Science, welcomed researchers, academic staff and postgraduate students to TU Dublin's Grangegorman campus on Thursday 3 September 2026 for its first Seminar on Explainable AI, a showcase of the breadth and ambition of current XAI research within the School.
The Centre is the first of its kind in the Republic of Ireland, dedicated to advancing transparent, ethical and interdisciplinary AI across domains including healthcare, neuroscience, law, energy and finance. This inaugural seminar brought its researchers together to share work on making AI systems more explainable, trustworthy and human-centred, and to open a wider conversation across the School's research community. Following a morning of presentations and discussion over coffee, the event was warmly received as the first in what the Centre intends to be an ongoing series.
Following a welcome from Dr Bujar Raufi and Dr Lucas Rizzo of the XAI Centre, the seminar featured five presentations spanning the field:
Why AI-Generated Text Detection Fails: Evidence from Explainable AI Beyond Benchmark Accuracy — presented by Sushanta Padasaini, examining why detectors of AI-generated text underperform once explainability is brought to bear, beyond their headline benchmark scores.
Design as Transformation: A Designer-Led Approach to Human-Centred Explainable AI — presented by Helen Sheridan, exploring how design practice can place people at the centre of how AI systems are explained and understood.
WiFi-based Human Activity Recognition — presented by Amany Elkelany, on recognising human activity through WiFi signals and the role of explainability in interpreting such models.
Exploring the Accuracy–Size Trade-off in Extended Argumentative Decision Graphs Using SHAP and LIME Feature Selection — presented by Dr Lucas Rizzo, on balancing model performance against complexity in argumentative decision graphs.
Explainable EEG Microstate Classification for Cognitive Workload Estimation: A Multi-Method XAI Analysis Using SHAP, LIME, DiCE and Anchors — presented by Dr Bujar Raufi, applying multiple explainability methods to the classification of brain-signal microstates for estimating cognitive workload.
The session closed with an open discussion on the direction of explainable AI research and opportunities for collaboration across the School and beyond.
Acknowledgements
The Seminar on Explainable AI marks the beginning of a new forum for the School's research community to share ideas, build collaboration and advance the understanding of transparent and trustworthy AI. The School of Computer Science extends its thanks to the Centre of eXplainable Artificial Intelligence, to all who presented, and to everyone who attended and contributed to the discussion.
The Centre of eXplainable Artificial Intelligence welcomes researchers, students and industry partners with an interest in XAI. To learn more about its work, visit the Centre's page on the TU Dublin website.