Hello Andrea, it's a pleasure to meet you. Could you start by briefly introducing yourself and sharing your experience in the field of artificial intelligence?
Andrea: Hello, I'm Andrea, an Associate Professor at the Department of Engineering and Information Science at the University of Trento. I've always been interested in hybridizing various themes in the field of artificial intelligence, such as machine learning and automated reasoning, and more recently artificial intelligence and human reasoning.

I am the coordinator of the Structured Machine Learning Group, where we deal with artificial intelligence models for domains characterized by complex structures, ranging from theoretical topics of academic interest to high-impact application use cases, especially in recent times in the healthcare field. For a few months now, I have been the coordinator of a large European research project called TANGO.

What is your specific role as the coordinator of the TANGO project, and what are the key responsibilities it entails?

TANGO is a highly interdisciplinary project, spanning from machine learning, artificial intelligence, human-computer interaction, decision making, to cognitive sciences. The real challenge of coordinating such a project is to bring together these components and guide them to produce a new generation of technologies that truly serve the individual. The idea is to develop decision support systems that enhance the capabilities of individuals and groups in decision making, such as identifying the correct diagnosis and the most appropriate treatment for a specific patient, or choosing the most effective actions to improve the conditions of a disadvantaged segment of society. TANGO is also a complex project from a management perspective, with 22 partners from 9 different countries, and a total budget of around 8 million euros for a duration of 4 years. Certainly, a great challenge that marks an important milestone in my professional career.

The TANGO project is coordinated by the University of Trento. Could you explain the role and importance of the University of Trento in the AI landscape?

The University of Trento has a long track record of excellent research in AI, with topics ranging from automated reasoning and knowledge representation to computer vision and machine learning. Excellence in both machine learning and deep learning research and automated reasoning is a distinctive feature of the University of Trento, making it the ideal institution to coordinate such an ambitious R&D project. We can confidently say that the University, particularly the Department of Engineering and Information Science (DISI), is a hub of excellence, both at the Italian and European levels, for AI. This is due to a series of historical reasons, including the proximity to the Bruno Kessler Foundation (FBK), with which there are numerous scientific-technological collaborations and joint activities. The theme of AI is clearly interdisciplinary, involving not only those, like me and my research group, who deal with the more scientific and technological aspects but also the social sciences in various aspects. The University is currently active on various fronts relevant to the AI ​​theme. There is clearly the research theme, which is what engages me the most, but there is also that of education (for example, we have a master’s program in Artificial Intelligence Systems active at DISI) and last but not least, technology transfer and the establishment of collaborations with local stakeholders, both public and private.

What are the main objectives of the University of Trento in the context of artificial intelligence, and how do these objectives align with the TANGO project?

In line with the European Commission’s approach to AI research, the objectives of the University of Trento are to create an AI that is genuinely human-centric and reliable, meaning that it places the individual, her needs, and society as a whole at the center and enhances their potential, respecting individual characteristics. A “good” AI that knows how to help people by “multiplying” their skills, knowledge, and strengths. Ultimately providing a kind of “superpower” - always leaving control in the hands of the individual, in line with an ethical and responsible approach to technology. In this context, the TANGO project aims to develop the theoretical, methodological, and practical tools to achieve this vision.

How did the idea behind the TANGO project originate, and how do you envision the collaboration among the 21 partners from nine European countries contributing to the project’s success?

The idea stemmed from the recognition that while much was being done in the field of AI to enable machines to explain their decisions (the so-called explainable AI), a true synergy between humans and AI required mutual understanding, where both parties learn to understand each other and evolve to bridge their respective gaps, achieving results together that wouldn’t be attainable without the contribution of each. Hence the name of the project, TANGO, because ‘it takes two to tango…’. Achieving such an objective requires a highly interdisciplinary effort. For this reason, I built a very broad consortium covering expertise ranging from machine learning and automated reasoning to cognitive sciences and human-computer interaction. It also included experts and stakeholders in various applied fields such as cardiovascular surgery, gynecology, banking credit, and public policies, and encompassed diversity (including - but not limited to - diversity among the member states involved), which has been a defining factor of the European Union since its inception.

The TANGO project already has an end-user here in Trentino, the regional healthcare service provider (APSS), which will test the developed technologies in a real-world context. How do you feel about this pilot project?

I consider it an excellent case study for the project. We’re talking about decisions made by surgical teams, which are highly complex and critical for patient health, where the amount of information to manage is enormous, and lack of coordination is responsible for most of the errors that occur. Creating an agent that monitors the operation and provides suggestions on decisions to make and warnings about anomalous situations that could lead to problems has enormous potential in terms of supporting the success of surgical operations and preventing complications, and more generally, in supporting medical decision-making."

Could you share some of the main challenges that human-centric AI must face and how do you think the TANGO project can contribute to addressing them?

One of the main challenges to address is indeed how to enable a form of mutual understanding between machines and people, which is a huge challenge given the great diversity that exists among individuals and AI systems. While knowledge on enabling AI to explain its functioning to users has accumulated over the years, the reverse (i.e., explaining to AI why we make certain decisions or take certain actions) remains largely unexplore. The project aims to overcome this challenge by combining two worlds of research, which are very distant in terms of history and approaches, namely cognitive and computer sciences, to create communication protocols that are aware of how people and machines reason and find a common ground where diversity is not an obstacle to understanding but a source of enrichment for the final result.

What is your vision of the future of human-centric artificial intelligence? What could be its most promising applications?

I believe we are at a turning point in artificial intelligence research, and perhaps also in how it can change many relevant aspects of our communities and societies. On the one hand, the ‘wrong’ (or malicious) use of artificial intelligence can encourage the spread of fake news or control over people, but on the other hand, truly human-centric artificial intelligence can revolutionize how we approach problems and can exponentially extend our ability to tackle them successfully. Imagine a system that can support you in your everyday decisions, with vast knowledge (because, simplifying, it has read everything that has ever been written on every single subject, it’s essentially a true polymath) and can help you understand the potential effects of each option, always available and updated. And it knows your preferences and habits, so it can truly personalize the suggestions it can give you. The most obvious applications are those related to decision-making, in medical, personal, or business contexts, but also in the development of effective and non-discriminatory public policies. Overall, such AI has the potential to help us make the world much better than we could on our own.

Finally, what are your expectations for the success of the TANGO project and how do you think it can influence the landscape of AI in Europe and particularly in Trentino?

I am very optimistic; at the kickoff, there were over sixty people representing the 22 partners, a level of participation that is rarely seen in projects and demonstrates the level of involvement of the partners and their desire to contribute to meeting this challenge. We already have several expressions of interest from other groups and projects at European and non-European levels, and we believe that TANGO can help pave the way to make AI genuinely human-centric.

All that remains is to stay tuned to discover the project’s developments. Thank you, Andrea, for your time! And good luck to the project!