António Ferreira is a Senior Researcher at the Research Centre for Territory, Transports and Environment of the University of Porto and a Lecturer / Assistant Professor at the Department of Civil Engineering and Geo-resources of the Faculty of Engineering of the University of Porto. Currently, his key research interests are focused on child-friendly futures and post-growth planning, the governance of the digital transition, and the governance of “smart” cities. António is also a Yoga and Meditation teacher and Personal Trainer and actively uses insights from these embodied disciplines to inspire his work. He is the Member n. 218 of the Portuguese Association of Urbanists.

DataColab and Generative AI: Exploring Large Language Models for Collaborative Data Analysis                                                                                                                                                                     Abstract: Turning complex data into actionable knowledge requires both technical expertise and effective collaboration. This presentation explores how DataColab could integrate large language models (LLMs) to support more accessible, interactive, and transparent data analysis. We propose a framework in which generative language models assist users in formulating research questions, exploring datasets through natural language, generating analytical code, and communicating findings. Within this framework, DataColab serves as a shared environment where researchers, analysts, and domain experts can review AI-generated outputs and collaboratively refine interpretations. Particular attention is given to grounding responses in relevant data and documentation, preserving the provenance of analytical steps, and incorporating human review throughout the workflow. The presentation also examines challenges that must be addressed before such capabilities can be used reliably, including inaccurate model outputs, data privacy, bias, and reproducibility. An evaluation approach is proposed to assess analytical correctness, usability, time savings, and the ability of users to identify and correct errors. By connecting conversational interfaces with collaborative analytical practices, this work outlines a practical direction for integrating generative AI into DataColab. It positions LLMs as assistants whose contributions remain subject to evidence, verification, and human judgment.