The increased number of government investments and the ongoing debate among European and international institutions about AI regulations, shows how Artificial Intelligence represents a strategic field not only for private companies but also for countries’ governments. With this respect, Europe has a significant delay compared to the United States and China: once again, many members states struggle to find a common and agreeable strategy and it thus seems that they are more focused on reaching consensus on normative matters rather than on financial and technological ones. While French and Germany seem to have reached a common solution, Italy is still in the process of defining strategy’s content. At the end of 2021, the national plan was approved, and the Government announced the creation of a National Fund aimed to boost and implement AI in order to aggregate resources and endorse innovative research.
Artificial Intelligence in Italy: a step forward investments
In a fragmented and diverse scenario, it seems hard for Germany, France, and Italy to reach a common AI solution. Furthermore, there is little doubt that Berlin, Paris, and Rome consider AI as a genuine national interest to defend and promote. Looking at the big picture, the development of this new technology is processing at different levels based on strategic choices following the traditional and deeply rooted legacy of our country’s economy.
AI strategies and policies in France
Paris, along with policy makers, is developing its own AI system to safeguard national language. ChatGPT’s launch in November 2022 kicked off AI strategy with subsequent efforts to implement a national sample: it is about Mistral AI, a brand-new start up founded last June which, in few weeks, got 105 million euros and which, last September, released the first version of its 7B language model. Through this language, including 7 billion different parameters, Mistral AI aims at adopting the new frontier of generative AI models providing users with an alternative open source to the current solutions. One of the main benefits in resorting to its models result in their superiority in terms of adaptability allowing customization for users’ specific tasks and needs. This adaptability not only helps companies to keep costs lower, but it also helps them face potential ethical challenges.
Mistral AI staff members are mostly data scientists, software, and machine learning engineers of leading companies like DeepMind, Meta and Hugging Face, whose main goal is to develop and implement AI models able to exceed OpenAI by the end of the year. From a financial perspective, Lightspeed Venture Partner has led the fundraising attracting dozens of investors, like Eric Schmidt, Google former CEO, and Exor Ventures, venture capital of Agnelli-Elkann holding company.
Europe’s first independent research lab dedicated to AI open science
Last November, the French startup achieved another milestone with the foundation of Kyutai, the first European non-profit research lab dedicated to open research in artificial intelligence and whose mission is to tackle the main challenges posed by the modern AI being committed to its democratization. The lab is strongly endorsed by Iliad, French TLC mobile operator, whose founder, Xavier Niel, shares this project together with Rodolphe Saade’, president of CMA CGM logistic group, and Eric Schimdt. Kyutai currently holds 300 million euros: 100 euros from Iliad, 100 euros from CMA CGM and other 100 euros from Eirc Schmidt foundation. Other important partners are Nvidia co-founder and CEO, Jen-Hsun Huang, providing the lab with the most innovative technological solutions, and Iliad Scaleway, specialized in cloud computing, providing it with high tech computing and GPU H100. Kyutai is currently recruiting new researchers and scientists committed to AI studies, specifically to developing large multimodal models and democratizing AI to share progresses with academics, scientific community, developers and companies. Its greatest ambition is to train as many AI talents as possible to throw down the gauntlet to AI giants: China and United States.
Germany to invest 1.6 billion euros in AI in the next couple of years
Berlin has recently presented its strategy to implement AI and it has allocated 1.6 billion euros for the next couple of years. Available resources will be distributed in 12 different areas with 50 different measures aimed at strengthening AI tools. Concerning scientific research, Berlin is planning on founding 150 academic labs, increasing the number of data centers all over Germany and making dataset accessible and available to implement AIs. As stated by the German Minister of Research Bettina Stark-Watzinger: “the only way to keep pace with the United States and China, is to work closely with our European partners. This is the reason why we want to strengthen European cooperation in AI field and optimize Europe’s impact”.
German AI startups increased by 67% in 2022
According to the German AI Startup Landscape 2023 study conducted by AppliedAI Institute for Europe, 508 AI startups were founded in Germany (+67% compared to 2022): the number of startups working in the field doubled in the first months of the year, placing Germany at 9th place at global level. Furthermore, as shown by 2023 figures, AI startups are more resilient than the standard ones. There are still few noteworthy German projects: Celonis, Aleph Alpha and DeepL.
Celonis, headquartered in Munich and founded in 2011, is trying to develop a new kind of AI tools to improve business operations. Alep Alpha, Heidelberg start up, is specialized in language models for the administrative and industrial sector. Baden- Württemberg government administration is still resorting to those models, and it has already been financed by multiple investors, like IPAI, Bosch, SAP, for a value of 500 million dollars. Lastly DeepL, based in Colonia, has developed an efficient translation service, like Google Translate, being considered as the so-called unicorn start-up registering more than 1 billion dollars in January 2023.
German Federal Government endorsing AI through NITD
Berlin is promoting a public initiative aimed at supporting innovations related to Artificial Intelligence: it’s a federally driven effort called NITD, National Initiative for AI-based Transformation into the Data Economy. Through this initiative, multiple machine learning and deep learning projects will come together to improve data networking quality, to develop new standardized tests and to promote AI companies all over Germany. This initiative was launched by Acatech, National Academy of Science and Engineering, headquartered in Munich, and supported by 32 million euros of public funding. The NITD is the pivotal project to launch the digital strategy as it strives to unified platform for AI application by the end of 2025, provided they are user friendly and help companies better perform their activities.
Italian market increased by 32% in 2022
Undoubtedly, Italy is the least progressive country in the field of AI technology. According to figures from Artificial Intelligence Observatory of Politecnico di Milano, 2022 was a good year for our country: AI market reached 500 million euros, with a remarkable increase of 32% in a single year, of which 73% funded by Italian companies (365 million euros) while export projects accounted for the remaining 27% (135 million euros). Despite these encouraging figures, the level of investments is still lower if compared to those registered in Germany. While in Germany and France there are, respectively, 8 and 7 unicorn AI startups (33 in the United Kingdom), in Italy there is no one. There are currently 260 companies providing AI innovative solutions: 55% of them provide users and companies with solutions in the healthcare, marketing, finance, and cybersecurity field; 25% working on different solutions, and 15% including system integrators and consulting companies. According to the above-mentioned figures, “53% of Italian medium and large companies have started working on AI projects. Fields registering the highest number of AI projects are the manufacturing one (22%), finance-banking (16%) insurance (10%). In terms of applications, the main solutions offered by AI technologies in the Italian market are:
- Intelligence Data Processing (33%);
- Natural Language Processing (18%);
- AI Recommendation systems (18%);
- Robotic Process Automation (RPA), Chatbot/Virtual assistants and Computer Vision (31%).
The Italian Investment Fund for Artificial Intelligence development
The Italian Government has taken preventive measures, involving two main directives, to keep Italy up with other countries: the first one led, at the end of 2021, to the approval of the Strategic Program for Artificial Intelligence 2022-2024. In line with the European Strategy, this Program outlines twenty-four policies to be implemented over the next three years to implement AI in Italy through the creation and enhancement of skills, research, development programs and applications. The main goal of these directives is to improve AI technologies and making the most of them in Italy by identifying European and national investment funds. These initiatives, designed to improve talents and skills, aim to increase the number of available PhDs attracting lots of Italian researchers both for basic and applied research. The strategic program includes policies aimed to promote trainings and jobs in STEM, to foster expertise in the digital and AI field, to strengthen research ecosystem in our country thus encouraging the cooperation between the academics, scientists, public bodies, society, and industry. Another important objective is to promote projects encouraging AI experts and professionals to return to our country, to finance platforms to sharing data and software at national level. Finally, the program aims at implementing policies leveraging AI both in daily business and public administration improving customer service and experience. It entails innovative measures within companies aimed at encouraging the 4.0 transition, creating innovative AI companies while testing and auditing new tools.
Measures in the public administration strive to create data infrastructures to safely use the potential of big data generated by PA, to customize public services and innovate administrations by strengthening GovTech ecosystems. This measure, for instance, entails the introduction of recurring announcements to identify and support brand-new startups providing them with AI solutions able to improve the public sector. In addition to this theoretical phase a practical one has been defined resulting in the announcement of the first Italian public fund for the artificial intelligence. The fund will be made available in partnership with Cassa Depositi e Prestiti and will have a budget of 200 million euros. The fund will be also managed by Cassa Depositi e Prestiti Venture Capital Sgr, holding the Innovation National Fund, with the support of an investments’ committee and advisory board. The Department for the Digital Transformation, under the Presidency of the Council of Ministers, will contribute with 45 million euros, while the Agency for National Cybersecurity will invest 40 million euros: the goal is to raise an additional 115 million euros from companies working in the field and from institutions like Cdp itself. This significant amount of money should be multiplied three times to be used for private investments reaching the 600 million target. This way, the fund would acquire startups, specialized funds, share, accelerators, and incubators’ shares.
Italian Government centralizes Artificial Intelligence
Last February, the Italian Government approved a legislative decree including provisions regarding the allocation of 150 million euros to support venture capital’s funds focused on AI, quantum computing, cybersecurity, 5G networks and telecommunications. More specifically, Chigi Palace authorized the Department for Digital Transformation and the Agency for National Cyber Security (ACN) to “jointly invest in shares and stocks of multiple venture capital funds, created by Cdp venture capital.”
Later this year, on April 8, the Government drafted the Artificial Intelligence law aimed at improving the management and use of AI resources; the Law provides a framework for AI decision-making processes and sets limits adhering to ethical principles of respect for human autonomy and prevention of harm. Furthermore, it highlights how important it is to safeguard the institutional and political democracy and prevent them from being hacked and manipulated during elections through the so-called deepfakes. The Department for the Digital Innovation will release the AI National Strategy based on guidelines provided by a committee of experts, incorporating specific goals which will be updated and revised every two years. The Department itself will make sure that the strategy is properly put into effect while the management of a proper AI use will be entrusted to government agencies, like Agenzia per l’Italia Digitale (AgID) and the Agency for the Cybersecurity (ACN).
France, Germany and Italy: new AI regulation agreement
France, Germany, and Italy are on the same wavelength about AI regulation at European level. These countries have, indeed, reached a common agreement finalizing the law proposal, known as AI Act, presented two years ago by the European Commission. More specifically, the agreement provides developers and deployers with clear requirements and mandatory regulations to be put into force by resorting to the code of ethics. However, it can be only applied to AI applications and not to other technologies.
On the one hand, developers should define a model card to provide users with detailed information on how models work while on the other a European authority should make sure that everyone is compliant with the code of ethics which includes 11 bullet points for companies to be as compliant and clear as possible while developing AI systems and Large Language Models. The Code encourages companies to employ suitable measures for identifying, evaluating, and mitigating risks of AI applications while also finding innovative solutions to potential incidents and misuse of emerging technologies. Beyond investing in security controls, companies should also publish reports on AI systems’ limitations, usage, capabilities, and abuse.
Initially, no one will be subject to sanctions, but they will then be applied after a certain timeframe. While the European Parliament has proposed to make the code binding only for majors AI providers, other countries have agreed on making it applicable to all providers.
Final thoughts
The current framework highlights a huge delay for Italy in terms of AI implementation tools and the greatest risk is to be left behind the main international players in such a key segment for our technological future. It is hard to compete with American and China tech giants having plenty of available resources for the development and improvement of inhouse AI technologies; it is nonetheless bad that Italy has just started to take the first steps for implementing its AI strategy. This is what happened for the national plan approved in 2021 and for the adoption of measures to implement private and public investments. Government’s national fund, aimed to aggregate as many financial resources as possible and to encourage startups’ foundation, is still a utopia. As it happened in France and Germany, the Italian Government should have played a more effective and forward-thinking approach to AI strategy but there is still hope with the national plan approved by Draghi.
