STOA Delegation to the ELLIS Unit in Milan

Written by Philip Boucher

Three members of the Panel for the Future of Science and Technology (STOA) – Patrizia TOIA (S&D, Italy), Maria-Manuel LEITÃO-MARQUES (S&D, Portugal) and Pietro FIOCCHI (ECR, Italy) – participated in a STOA delegation to the European Laboratory for Learning and Intelligent Systems (ELLIS) Unit in Milan on 23 and 24 May 2022. They met with researchers working on various fields of artificial intelligence (AI) including technical development, assessment techniques, industry collaboration, ethics and policy.

The European Laboratory for Learning and Intelligent Systems (ELLIS) is a pan-European AI network of excellence that focuses on fundamental science, technical innovation and societal impact. It has recently established a Unit in Milan, which aims to develop into a key European hub for AI by promoting scientific excellence and fostering top-quality industrial research. It coordinates and focuses the AI research of four local universities: University of Milan, Bocconi University, Politecnico di Milano, and University of Milan-Bicocca. These Universities take turns to chair the Milan ELLIS Unit. This year, the University of Milan, our hosts for the delegation, has the chair. The Unit’s work is particularly relevant in the context of several legislative proposals, including the AI act, data governance act, digital services act and the regulation on machinery products.

The delegation opened with a visit to the Rectorate of the University of Milan. Members were welcomed by the Rector, Professor Elio Franzini, Deputy Rector Professor Maria Pia Abbracchio and Professor Nicolò Cesa‑Bianchi. They provided an introduction to ELLIS and the Milan Unit and participated in a discussion about the modalities of industrial collaboration, including sponsorship, capacity building, and the management of intellectual property rights.

The second day of the delegation was dedicated to in-depth meetings with researchers from the ELLIS Unit. Members visited the Department of Computer Science at the University of Milan and were welcomed by its Director, Professor Silvana Castano, who highlighted several collaborative research and study activities. The day continued with two sessions of meetings with ELLIS Unit researchers from across the four universities.

Opening the first session, Professor Vincenzo Piuri introduced several projects applying AI to monitor and control industrial processes, renewable energy production, transportation systems and security. Professor Andrea Celli then explained his work on mechanisms to ensure the fairness of ‘matching platforms’ that connect users and services, for example in the distribution of online advertising. The session continued with Professor Federico Cabitza, who highlighted the diminishing role of accuracy metrics in assessing the quality of machine learning applications in the medical domain, and set out alternatives that focus upon robustness, reliability and utility. Professor Viola Schiaffonati concluded the session with a discussion of how AI presents challenges to traditional conceptualisations of risk and uncertainty, calling for a paradigm shift to make discussions of AI ethics more effective, and highlighting the concepts of active responsibility and value sensitive design.

The second session opened with Professor Francesco Trovò, who highlighted the uneven application and potential of AI across industrial sectors, and explained several initiatives to promote uptake and technology transfer, including their key enabling and constraining features. Professor Simone Melzi then presented the state of the art in image recognition including cutting-edge developments with 3D data, highlighting the challenges presented by image manipulation and adversarial attacks. The delegation then heard from Professor Carlo Baldassi, who explained the potential benefits of recent advances in the field of computational neuroscience, including systems that interact directly with human brains, and their associated practical, theoretical, and ethics challenges. Finally, Dr Massimo Rivolta showcased ELLIS work on AI decision support systems for medical diagnosis and treatment, highlighting key elements related to data availability, trust, privacy and personalisation.

The delegation concluded with a visit to Fondazione UNIMI, the university’s start-up incubator. Members were welcomed by the incubator’s President, Professor Luca Solari, and its Director, Dr Roberto Tiezzi. They explained the Fondazione’s model for bringing academic research to the market applications, and engaged in discussion of the associated opportunities and challenges. They were then joined by Professor Caterina La Porta, who presented several innovative projects in the field of pathology, including inspiring applications for cancer diagnosis and treatment. The discussion focused upon the need for good-quality, interoperable health data, a key factor in the European Health Data Space.

STOA is grateful to the universities and staff for their warm welcome, the detailed explanations of their work, and the stimulating discussions. We look forward to further opportunities to engage with universities and researchers in the future.