Artificial intelligence - interview with Professor Michela Milano, director of the ALMA AI Centre

Artificial intelligence

Artificial intelligence is not only increasingly impacting our daily lives, but it is also revolutionising research.

The Alma Mater Research Institute for Human-Centred Artificial Intelligence (Alma AI) is the University of Bologna’s interdisciplinary solution to the major challenges linked to artificial intelligence, with the aim of bringing together and enhancing the competencies of researchers working in the University’s various departments.  

Now that a year has passed since it was set up, we recently spoke to Professor Michela Milano, Director of the Centre, to take stock and gain a deeper understanding of what the future holds in terms of artificial intelligence and the impact and benefits it will have on society.

Artificial intelligenceWhat was the initial aim of ALMA AI?

ALMA AI is enabling us to bring together, in a single centre, more than 500 teachers, researchers, PhD students and research fellows with competencies and active research in AI in order to build connections and synergies, integrate knowledge, and promoting dialogue between those involved through internal networking. Artificial intelligence (AI) is playing an increasingly important role in many aspects of our daily lives: in machines, in industry, in the world of work and in the public sector. However, from a research point of view, there is still a lot to be done.

The centre will allow us to improve our university’s performance when it comes to developing projects and seizing funding opportunities on several levels (regional, national and European), collaborating with businesses, and third mission goals for the local area. In addition to being a research centre, Alma AI also has dissemination goals. As a generalist university, we are able to branch out and see how AI can be applied in different fields, establishing internal relationships to improve the university’s performance.

Research Institute for Human-Centered Artificial Intelligence

The ALMA AI project originated in Bologna. What potential does this area have to offer?

We live in a city that serves almost as a natural cradle for artificial intelligence, backed by about 30 years of teaching and research into AI, but not just that: we also have an ecosystem, a significant infrastructure. AI requires a lot of computing resources, and Bologna accounts for more or less 70% of Italian supercomputing. Cineca and INFN are two key centres in this infrastructure, hosting data centres containing a substantial amount of computing and storage power, and therefore memory. Another important project is the Tecnopolo, which is being developed in Bologna, and BI-REX, a Competence Centre in the Industry 4.0 funded by the Ministry of Economic Development, which will serve as a hub for Emilia-Romagna’s industrial businesses along the lines of Europe’s digital innovation hubs. Moreover, AI is also playing a crucial role at regional level: the regional Big Data for Human Development Foundation (IFAB) was recently set up, which brings together the region’s universities, research centres such as ENEA, CMCC, CNR, INGV, IOR and other local players active in supercomputing such as Cineca and INFN.

How is the centre structured to combine these research areas?

To make the most of the wealth of competencies available in our university, eight scientific units representing the different areas of expertise have been set up:

  • Foundations of AI and AI e Hard Sciences cover the key elements of the discipline: computer science, from which AI originates, the algorithmic part (how to build new technologies, new algorithms and new tools that allow us to improve research and applications), and hard sciences, i.e. mathematics and physics, for machine learning (deep learning), large mathematical models, particle physics (also known as high energy physics), simulations, etc.
  • Humanistic AI, which deals with the links between AI and the humanities (for example, the use of AI for understanding natural language and its categorisation, etc.).
  • AI for industry, dedicated to industrial applications and the uses of AI within companies, where AI comes into play in logistics management, recruitment support systems, and in systems for optimising production and maintenance operations in companies.
  • AI and High Performance Computing, focusing on the analysis of hardware platforms to improve AI algorithms and how AI can help improve energy efficiency both in the industry and for supercomputing, thus reducing energy consumption.
  • AI for Health and Well-being linked to medicine to improve processes and medical imaging, assess patients’ risk factors based on clinical examinations, and even so-called personalised medicine: a new frontier in the development of non-standardised treatments, which are targeted to each patient. In Bologna, we collaborate with the three IRCSS - Scientific Institute for hospitalisation and health care (Sant’Orsola Hospital, Bellaria and Rizzoli) to develop AI applications in this field.
  • AI for Law and Governance, which originates from the former CIRSFID - Centre for Research in the History, Philosophy and Sociology of Law and in Computer Science and Law in order to develop AI techniques in the legal and political-social field, understand how certain behaviours comply with laws, and assess the impact of the introduction of new regulations so as to avoid conflicts between regulations issued on the same subject.
  • AI and Education for applying AI to e-learning platforms and innovative teaching methods in general, as well as to AI learning processes, which is essential as the university’s primary goal is to provide education.  

What initiatives are there to link and make the many people of ALMA AI work together towards common research goals?

Despite the pandemic, we have started promoting WednesdAI – educational aperitifs with AI and thematic workshops where interdisciplinarity is a crucial aspect. The aim is to get to know each other and the huge university community involved in AI. Getting all these people to communicate is not easy, but it is worth it: we often realise that we speak completely different languages, even though we are talking about the same concept. Eventually, they will become one: as soon as they merge, we can expect the result to be greater than the sum of its parts.

Can you tell us something about the activities, initiatives and significant results achieved one year after the launch of ALMA AI? 

The first goal is to establish a link between the centre and the main AI-related institutions in the local area. We are entering into an agreement with the INFN - National Institute for Nuclear Physics in the area of high-performance computing, artificial intelligence and hard sciences, which will also involve sharing their computing infrastructure. We are also establishing collaborations with Cineca and the CNR - National Research Council.

In the medical field, an agreement has been signed to set up a joint laboratory with the IOR - Rizzoli Orthopaedic Institute dedicated to research on medical data and how it can be used while respecting patients’ privacy. Another agreement was formalised with the AUSL (local health authority) of the Sant’Orsola Hospital for studies on logistics, an area in which AI can improve or support healthcare decision-makers to better organise their activities.

Another current research theme is that of federated approaches to machine learning so as not to move data, but only data algorithms.

Another initiative that I would like to highlight is related to the co-innovation lab: we are building a laboratory equipped with a computing infrastructure donated or loaned for use by several hardware companies. Our researchers and companies can access this project to develop joint projects.  The university therefore has access to new cutting-edge equipment at no extra cost and without being tied to a single supplier, while the various companies that supply the hardware not only get to use and test it with our researchers, but also have the opportunity to publicise and promote their tools.

What initiatives is ALMA AI involved in at European and international level?

The Centre is active in a number of international initiatives such as the one launched by the EU in 2019 to develop the AI4EU platform (AI on demand platform), an aggregator of all European AI resources. Our university also coordinates one of the follow-up projects to improve and simplify platform use for non-specialist users (StairwAI project), which will allow a multilingual interface to be built through which users will be able to “speak” directly to the platform in their mother tongue.

The ALMA AI centre is also involved in the Global Partnership on Artificial Intelligence, a global initiative on the social implications and impact that AI can have on the world of work, on ethics and on our daily lives.

The centre also boasts seven winners of AI-related ERC projects among its members. The ERC projects of excellence, which are funded by the European Research Council, are the strong point of our centre.

Can you give us some examples where our university’s AI research team has developed excellence and helped generate tangible benefits for society (communities, local businesses or civil society)?

AI is based on building three types of models from data:

  • descriptive models to create representations of the reality of a system you want to work on (e.g. a plant, a city, an electricity grid or a natural ecosystem) describing its components, interactions and how the various processes work;
  • predictive models, which attempt to predict the future dynamics of the systems being studied;
  • decision-making models, which help decision-makers without taking their place. These models, for instance, can help a political decision-maker to choose a mobility plan for a specific district or to define strategic energy plans for a specific region. In the same way, they can also be provided to a business manager to improve logistics, staff shifts, or to control production lines in order to improve the process and the product.

AI also offers a much broader range of applications: for instance, in the field of social networks, with techniques that allow natural language to be analysed to find out what a specific audience thinks about a certain subject.

Further applications include the use of AI in autonomous driving so that cars can recognise obstacles ahead, thus improving their performance – something that will become increasingly common on our roads.

The advent of AI has also raised a few concerns among discerning observers: on the one hand, the malicious uses of this technology and, on the other hand, the impact on the world of work and the risk of causing more “technological unemployment”. What do you think?

It goes without saying that there are also possible uses of AI that we should be concerned about. The important thing is to study them and be aware of them. With AI, we can build images and videos that are very similar to existing works. We can also create artificial human faces that are absolutely identical and indistinguishable from physical ones. In this way, so-called “deep fakes” can be created: this malicious use needs to be studied and understood in order to recognise, also automatically, false images from real ones and remove false news from the Web.

Another area of considerable interest in the use of AI is the development of fully automatic weapons. A weapon system with AI, making it totally independent when it comes to deciding when and whom to hit, is extremely dangerous and is something that goes beyond all imagination. The international scientific community is aware of these risks.

AI brings with it a real risk of “technological unemployment”. A recent McKinsey report estimates that 30% of jobs will be lost to AI by 2030. In order to tackle these risks, workers need to be somehow prepared to build up competencies even throughout their professional lives, and not just stick to what they learnt at university or school. In fact, they need to continuously keep up to date and learn so that they can interact with these new systems.

The European Regulation on AI has recently been introduced, which will regulate permitted and non-permitted uses of AI, focusing especially on high-risk applications. With its wealth of technological, scientific and legal competencies, the centre can act as a privileged partner to study and test this regulation.

Where will AI take us? Is it possible to come up with innovations that are still unimaginable in this field? 

Unimaginable things are already happening. For example, you may have heard the news about Google DeepMind, which has just solved the “protein folding” problem” that has been ongoing for more than 50 years. AI can improve work in numerous fields and can advance human progress in the field of sustainable development and in the achievement of the Sustainable Development Goals defined by the UN.

Another window on the unimaginable is the one that opens on so-called “general intelligence”, namely an AI closer to the human one. With AI, we can develop highly accurate techniques, in some cases even with superhuman abilities, such as computer vision. There is still a long way to go, but the direction is clear.

Back to today, what do you think are the main obstacles to the development of AI and how can Alma Mater’s research community prepare itself to overcome them in the future?

One of the main problems in general is that of data access and control. Data is one of the main enablers of AI, but it is often not available, or is of very poor quality or private. Privacy concerns are high in Europe, but this is not always the case in other parts of the world. The key issue is to find a way to use data (e.g. PA data, clinical and medical data, personal data, etc.) so that it can be used to extract value from it, without violating privacy.

Moreover, in Italy and Europe there is also a lack of funds as we are not yet in line with what is happening in North America or Asia. Guaranteed and continuous funding is essential for AI development, as is committing to reducing the fragmentation of the ecosystem in Italian and European research. Our centre was specifically set up with the aim of bringing together researchers within our community and our local area.

Published on: 04 June 2021