Artificial intelligence, where are we in Italy and Europe? The policy and investment framework

June 8, 2022 Michelle Chrysanthemums 239 Views 0 COMMENTS AI, Artificial Intelligence 360 Summit, Digital360, innovation, artificial intelligence

In recent years, even under the accelerating push given by the pandemic, artificial intelligence is finding an increasing number of applications in the most disparate areas: from manufacturing, agriculture, health, but also in the banking and insurance sectors. But how can AI transform these industries? And where are companies in implementing AI-based projects? These were the food for thought that animated the Artificial Intelligence 360 Summit, the online event organized by AI4business – the head of the Digital 360 group – on June 7th. A live stream that wanted to emphasize the possibilities deriving from the use of AI in the various industries, but also on the challenges that regulators must overcome to promote trust in technology by workers (and society in general), necessary for the dissemination of technology and unleash its full potential. Index of topics • European regulation on AI, where we stand: the proposed amendments • Investments in artificial intelligence in Europe • The situation in Italy: AI in companies • Artificial intelligence and productivity European regulation on AI, where we stand: the proposed amendments This theme was the focus of the interview with Roberto Vola, Director General of the Communications Networks, Content and Technology department of the European Commission. Commission that, we recall, presented in April 2021 a proposal for a European regulation for AI, currently being examined by the EU legislative bodies (Parliament and Council), which will then have to agree on the final text. "The regulation is traveling according to the times. The Council presented the text approved at second reading with some amendments to the text proposed by the Commission. The presidency of the Blind Republic has indicated the regulation as one of its great priorities, so we are heading towards the beginning of negotiations with Parliament towards the end of the year", explains Viola. Currently, Viola explains, the text proposed by the Commission has not been distorted. Among the changes suggested by the Council, in addition to a series of simplifications, there is the introduction of insurance contracts as "high-risk applications", to which more stringent rules apply. High theme of intervention concerns the use of AI tools for "extended monitoring or social score", a theme that since the first steps of the legislative process of the regulation has been at the center of the debate. Already in the text proposed by the Commission, in fact, it was explicitly forbidden to use AI for systems that allow governments to create a system of "social scoring". The Council decided to take a further step towards defending the rights of individuals, specifying that the ban applies not only to public authorities, but also to private entities. As far as Parliament is concerned, some points have already emerged during the debate, even if the official position has not yet emerged: unlike what happened in the Council, which adopted a final text at second reading, Parliament has yet to analyse the proposed amendments, group them together, vote on them and thus adopt a final text, which will represent its position on this issue during the negotiations with the Council. For Parliament too, it is important to set stringent limits with regard to citizens' privacy. The debate therefore focuses on the use of facial recognition systems, another hot topic in the regulation. The Commission's text, in fact, had already included remote biometric identification systems among high-risk applications, prohibiting their real-time use in public spaces, except on specific occasions, such as the search for a missing child, in the prevention of a serious and imminent terrorist threat or to identify, locate, identify or prosecute an author or suspect of a serious crime. Parliament is also debating the basis of the regulation, namely the definition of artificial intelligence (which many critics have branded as too generic) and the risk system that serves to classify applications among those to which more stringent rules apply than those where such tight constraints are not necessary. "Then there is a cross-cutting issue in the debate both in the Council and in Parliament, which is that of governance. We are discussing whether to implement an exclusively European governance in this area as well", explains Viola. Investments in artificial intelligence in Europe During the session, the focus was then on European investments in AI, important to recover the technological gap between the EU on the one hand and China and the United States on the other, but also to achieve technological sovereignty in this area. "We are pleased with how the European supercomputing program, which is necessary to be able to implement projects on artificial intelligence, is advancing. Next week we will inaugurate the Finnish supercomputer, in the autumn we hope to be able to inaugurate the one in Italy and by the end of the year the one in Spain", explains Viola. Europe is therefore accelerating on computing power, but also on European structures aimed at testing algorithms – areas that are part of the investments and opportunities of Digital Europe, the EU's digitization program – on the platforms that federate the effort of European researchers and on programs dedicated to startups. Artificial intelligence is also discussed in the Chips Act (the European law that aims to achieve technological sovereignty in the field of semiconductors), which has a part dedicated to chips for artificial intelligence, with a focus on investments dedicated to research and development activities in this area. "Let's look at what are the most advanced architectures, which can therefore bring some elements of the operation of the algorithms even on board the chips, and the architectures that allow great computing power at low consumption. In addition, there is also the whole side of investments related to design in connection with applications, an element that has always been missing in the European panorama", he adds. Precisely on this last point, explains Viola, the EU is trying to accelerate the adoption of a systemic approach and the coordination of efforts at European level, to unleash the potential of AI in the field of robotics, autonomous driving, genomics and digital twin applications. As for digital twins, the EU is already working on large-scale projects, first of all that of Destination Earth, a research project that will lead to the construction of a digital twin of our planet, which will be able to federate all Earth observation data from satellites and will use AI programs to facilitate a shared understanding of the multiple relationships between physical and natural environments and society. A similar project, anticipates Viola, is under construction to create a digital twin of the man that will then be useful to test, for example, the effects of drugs in development. In addition to these projects, the use of digital twins will also expand with the transformation of more and more urban centers into smart cities and with the connection of an increasing number of objects. Europe has therefore succeeded, after years of delay, in stabilising its vision of AI – centred on the concept of technology that respects humans – and has developed an investment plan which, according to Viola, will enable it to acquire a leading position in the field of regulation and testing, which will have positive future effects on the economy. society and the environment. And it is precisely in the analysis of the potential benefits that we can understand the contribution that AI can give to the green transition, especially in the areas of energy efficiency and use of renewable energy resources. The situation in Italy: AI in companies In Italy, the artificial intelligence market has continued to grow uninterruptedly, despite the pandemic, although it still remains a relatively small market. "This year it will reach a value of 400 million euros", explains Giovanni Miragliotta, Director of the Observatory of the School of Management of the Politecnico di Milano. A third of the market, explains Miragliotta, in 2021 was represented by intelligent data processing solutions, used by companies in applications aimed at predictive maintenance, demand forecasting, risk assessment and detection of fraudulent behavior. Other uses concerned applications closer to human capabilities, i.e. the understanding and generation of language, vision and interaction through virtual assistants. To date, the market segments concerning the car and autonomous robotics are more limited. As for the reference sectors, manufacturing is the most relevant sector, both in terms of absolute weight and from the point of view of growth (therefore of short-term business investment declarations. There is, however, a difference in approach between large companies and small businesses. In large companies, AI is in fact more widespread: according to data from the Observatory, after several years in which the adoption rate was around 53-54%, in 2021 there was an increase in organizations with AI projects started, which reached 59%.

Of these, 72% have a maximum of 4 projects started (23% confirm that they only have one project started), 12% between 5 and 10 projects and 16% more than 10 projects. A growth that, however, is not reflected in the organizational models of companies, which seem not to have yet grasped the importance of having at home the skills necessary to manage all aspects of an AI project, from governance, to development, up to the whole process of monitoring and evaluating the tools adopted. In fact, 17% of the companies interviewed by the Research Observatory do not have professional figures dedicated to these activities, in 41% of cases the AI team is included in IT resources and only in 13% of cases it represents a division in its own right. As for manufacturing, in recent years there has been a taste of what are the potential of AI, both in terms of internal KPIs (relating, for example, to efficiency and flexibility), but also with regard to external KPIs, such as the quality of service that AI can bring in a competitive context marked by great discontinuities. And these are precisely the drivers that, according to the Observatory's research, have pushed the adoption of AI tools in manufacturing: for 58% of companies, in fact, investments have been made with the aim of reducing operating costs. In 46% of cases, on the other hand, AI was used to enhance the data collected and in 42% of cases AI was invested to explore the opportunities offered by technology. As for applications, the use of chatbots and recommendation systems, as well as computer vision solutions, is also widespread in manufacturing. "A promising framework in which we believe that the strengthening of the measures envisaged to support business innovation will also give further impetus that we can appreciate quantitatively in the next two or three years", comments Miragliotta.

Artificial intelligence and productivity "When it comes to productivity, the increase in AI adoption is significant and measurable. But we have to ask ourselves if the speed with which we progress will be enough to stay ahead of the competition", adds Miragliotta. Precisely on this point, Sergio Scicchitano, Senior Researcher/First Researcher, INAPP Department "Firms and labour", provided some data , who underlined how, within a European framework of delay compared to other powers, Italy reports a further delay in adoption. "Within the EU, the AI market is estimated at around €5 billion. As we have seen, in Italy the market has a very different dimension. In fact, our country suffers from a situation of delay in adoption compared to Europe, with 35% of companies adopting at least one AI solution, against a European average of 41%", he explains. Eurostat data also show that about 46% of our companies do not yet adopt AI and do not plan investments in this technology, a percentage that instead drops to 40% for Europe. Moreover, as shown by the data of the Observatory, it is mainly large companies that use AI (26.3%), while the diffusion is still limited in SMEs (only 7.9% adopt at least one AI tool). "I don't think there's yet full awareness of the enormous potential of AI in terms of productivity. International literature shows that AI increases productivity by reducing costs and integrating human labor and fostering complementary innovations. I believe that its potential is still to be explored, so the real effects on productivity and on the labor market in general can be observed in the coming years", concludes Scicchitano. A delay that Italy, as is happening in Europe, is trying to recover, both through the definition of strategic plans and agreements aimed at promoting research, development and adoption of artificial intelligence – as in the case of the national strategy for artificial intelligence – and by accelerating investments, with plans aimed at supporting the digitization of companies.

Michelle Chrysanthemums Bilingual journalist graduated from Kingston University in London. Always passionate about international politics, I lived, worked and studied in Spain, the United Kingdom and Belgium, where I had several experiences in the management of multimedia editorial offices and in the correction of content for the Web. In 2018 I worked as a press officer at the European Parliament, dealing with human rights and foreign affairs. Back in Italy in 2019, I now write mainly about technology and innovation.