We are excited to share that our team will have a strong presence at this year’s Deep Learning Barcelona Symposium (DLBCN 2025), with several papers accepted across conferences in AI, human–computer interaction, and information retrieval. DLBCN is always an excellent opportunity to connect with the local research community, and this year it provides the perfect stage to showcase our recent advances related to interpretability, human-centred analysis, and user-centric evaluation, illustrating our commitment to combining behavioural data, neural signals, and machine learning to better understand how people interact with complex computational systems. This line of research directly informs SYMBIOTIK’s mission: designing information-visualization tools that make AI behaviour more transparent, more interpretable, and ultimately more aligned with human needs.
Specifically, our contributions span eye-tracking, cognitive signals, and user modelling, all of which offer new perspectives on how humans interact with modern computational tools. Below are the papers that Telefónica Scientific Research – UPC will be presenting as posters at DLBCN:
- Sebastian Idesis and Ángela López-Cardona: OASST-ETC Dataset: Alignment Signals from Eye-tracking Analysis of LLM Responses. Proc. ACM ETRA 2025.
- Ángela López-Cardona and Mireia Masias: Brain–Language Model Alignment: Insights into the Platonic Hypothesis and Intermediate-Layer Advantage. Proc. UniReps Workshop @ NeurIPS 2025.
- Georgios Koutroumpas: Beyond Clicks: Eye-Tracking Insights into User Responses to Different Recommendation Types. Proc. ACM RecSys 2025.
- Georgios Koutroumpas: Beyond One-Size-Fits-All: A Study of Neural and Behavioural Variability Across Recommendation Categories. Proc. ICTIR 2025.
- Ángela López-Cardona: Seeing Eye to AI: Human Alignment via Gaze-Based Response Rewards for Large Language Models. Proc. ICLR 2025 (oral presentation).
