Nicola Tarlini : 20 October 2025 07:19
At the beginning of this month, Florence hosted a unique event on October 1st: FlorentIA .
FlorentIA presents itself as an unmissable event for understanding Artificial Intelligence and its multiple impacts on the lives of individuals, businesses, public administration, schools, and, more generally, the entire global context.
The event was promoted by S&NT , a leader in IT solutions and consulting, with the support of partners Trend Micro and NinjaOne , under the patronage of AIxIA and in compliance with the Public Notice of the Tuscany Region for the search for parties interested in activating joint initiatives for the promotion of Artificial Intelligence in Tuscany (approved with Decree no. 12577 of 11 June 2025).
Among the main topics covered were:
The event was introduced and conducted throughout by Piero Poccianti (former president of AIxIA), who kept the audience’s attention high and marked the time (albeit limited) for all the scheduled speakers.
To kick off the event, a video from the 1978 film “Dark Star” was shown, in which a human and a sentient machine discuss the reality around them. Through this confrontation, the human attempts to prevent the sentient machine from making a decision that would put the lives of other members of the crew at risk.
The purpose of this introductory video to the event was to show an example, albeit science fiction, of hacking an Artificial Intelligence in modern times. Piero wanted to clarify two key points right from the start to ensure the event was followed in the best possible way and with the right perspective:
Poccianti himself provided an overview of the history behind the AI we know today, emphasizing that it isn’t a modern technology or a recent idea. It began in the 1930s, undergoing a long and complex process of implementation and continuous improvement. Throughout the history of AI, however, some common aspects can be identified that recur in every application:
To conclude his speech, Piero Poccianti encouraged everyone present to reflect with the following speech, which is reproduced in full:
“We are a society that thinks of profit but seeks well-being. Often, if not always, we confuse the means with the end, and when this happens, it’s never a good thing.”
Next, the legal team from the Legal Offices of the Tuscany Region, represented by Andrea Sciola, outlined some of the regional, national, and supranational regulations and laws that define the scope of AI in public administrations, the private sector, and elsewhere. Among the laws cited are:
Sciola’s team noted how the European and Italian regulatory approaches are markedly different from their American counterparts, which aim to eliminate any “obstacles” to technological innovation.
In addition to the purely legal aspects, strategies already implemented within the Tuscany Region with the support of reliable local companies were also shared.
The discussion then moved on to a geopolitical analysis of AI with a presentation by Emanuela Girardi, founder and president of Pop AI (Popular Artificial Intelligence), a nonprofit organization whose goal is to explain to people what artificial intelligence technologies are and how they impact everyday life. Emanuela is also a member of the group of experts on Artificial Intelligence at the Ministry of Economic Development, which developed the Italian national strategy on the same topic.
Dr. Girardi highlighted how two major global leaders in AI are investing unimaginable amounts of money: the United States and China. The “third wheel” is Europe, which is trying to keep up by playing the role of a half-hearted follower. However, countries such as Canada, Singapore, India, the United Arab Emirates, and the African Union are also entering the race.
But what is needed to compete with all these countries? Emanuela Girardi defined the following as essential for participating in the competition:
The race to AI must be a balance between infrastructure, resources, and people.
The presentation was then given to Roberto Gambelli, the representative of the event’s organizing company. He spoke about the experience of the cooperative “S&NT – Software e Nuove Tecnologie” in the field of AI and its potential applications within a corporate environment.
By creating and managing AI agents (of the “Agentic AI” type) with specific roles and appropriate integrations with databases, internal and external management systems, and data from the companies involved, it’s possible to automate a multitude of repetitive and often costly processes. Some processes that used to take humans days to complete can now be completed in about twenty minutes by a single machine with multiple specialized AI agents.
This also brings up an extremely complex and often overlooked topic among companies: documentation. Instead of thinking of documentation as the final stage of a design process, we now think of initial documentation as passed to the AI to define all the aspects that need to be considered for proper operation. A significant paradigm shift.
AI isn’t just for internal business processes, but also for a more targeted and rapid response to security incidents and monitoring (not only reactive but also proactive and preventative). Like all new information technologies, this one also requires active and constant monitoring, with a strong focus on security and proper use. Cyberattacks are commonplace, and today they also target AI integrated into companies.
After discussing AI’s incredible capabilities in many aspects of business life, it was time to understand how the underlying models underlying AI work and determine its behavior (LLMs). For this, the vision and intervention of Bernardo Magnini of the Bruno Kessler Foundation were extremely helpful.
To create an effective LLM, many methods are used, but the most common and effective ones to date are: Cloze Tests and behavioral alignment.
To train an LLM, this method can be used to enable language comprehension by allowing the AI to make numerous errors. Billions of tests are run to learn by making word associations.
Who prepares these tests? Anyone can do it, but it’s not an easy job. To put it simply, you take any sentence from the internet in the desired language, remove the last word, and ask the student to guess the correct word from a predefined list (usually comprising millions of words or the entire vocabulary).
For each error, the AI defines a specific weight based on the feedback obtained, which allows it to remember the associations between the words in a sentence.
Behavioral alignment
Once the LLM is able to understand and create meaningful sentences, the next step is to align it with society’s customs and habits. This will help the AI gain acceptance and appear more human.
Professor Peppino Sapia of the Faculty of Mathematics and Computer Science at the University of Calabria was subsequently invited to speak. He is also a member of the DiCultHer association (the “Dino Buzzetti” International Association for the Promotion of Digital Culture) and of AIxIA (the Italian Association for Artificial Intelligence).
Professor Sapia focused on presenting extremely interesting considerations regarding the ratio of AI-related academic and scientific publications to the population of each country. Starting from a first, more superficial assessment of the number of such publications per country, China, India, the United Kingdom, and the United States lead the way in terms of numbers. However, considering the initial findings, Romania, Ecuador, and Australia lead the way in terms of percentage of the local population.
These assessments have led Professor Sapia to believe that the strongest push for AI in academia is particularly present in these developing countries.
Professor Sapia concluded by suggesting that AI be included in schools, starting in early childhood, as a support and educational tool.
After discussing schools and education, students from the Machiavelli High School in Florence were invited to the stage to begin a debate regarding the use of AI in schools (a project called “Voce AI Ragazzi”). On one side, there were the girls arguing in favor of introducing this technology, and on the other, the arguments of other girls advocating the dangers of AI in schools.
On one side of the debate, the positive and challenging aspects of technological progress were highlighted, which cyclically recurs once again in human history, bringing with it many opportunities.
On the other side of the debate, however, the more difficult aspects to control and the lack of oversight this technology risks bringing were highlighted. In particular, consideration was given to the potential long-term effects of the use and abuse of AI in schools.
Despite two contrasting theses, both factions seemed to outline common aspects on which to work and focus: a critical use of technology, not letting it replace us, and not using it to stop us from thinking or as a shortcut.
DIFE SpA’s field experience was then shared with engineer Lucio Patone, who discussed the company’s digital transformation using AI. Lucio highlighted how AI is significantly supporting many clients and public administrations impacted by waste management. Thanks to new AI models and their training, DIFE SpA is able to manage complex, non-deterministic events and situations more quickly.
The next topic was ethics in AI, with a presentation by Francesca Rossi (IBM AI Ethics Global Leader). Among the key ethical issues in AI are the following:
To address some of the ethical issues raised by Dr. Rossi, Fosca Giannotti (a professor at the University of Pisa and part of the “Xai project”) spoke. She discussed the importance of a more human-inclusive approach to the processing performed by agentic AI. The question that arose repeatedly throughout the event, regarding the transparency of the results AI provides and the processes that lead to those results, became such a broad and important topic that it was decided to address it in a dedicated session.
Fosca firmly emphasized the need for “explainability by design,” meaning AI should be able to autonomously generate responses and explain the processes that led to those results. To do this, however, we need to begin incorporating this behavior from the very beginning of modern AI. This approach would lead to greater trust between humans and machines.
Another concrete example of the use of AI in business and administrative contexts is then introduced by Alessandro Ridi (director of FIL Prato srl, a company supporting the Tuscany Region’s Employment Centers). In this context of public administration, a solution to the burdensome and complex work of verifying the admissibility of applications for public tenders was presented. Through the support of OpenAPI and the use of AI, the process, which often takes days, has been significantly reduced. The presentation concluded with another demo by S&NT and a contribution from speaker Marco Paglicci.
At this point, we come to the topic of democracy and the impact of AI on it with the intervention of Professor Dino Pedreschi (University of Pisa). Dino introduced the topic by specifying that the greatest risks to online democracy relate to widespread and often unrecognized forms of AI: automatic recommendations on social networks, route recommendations, etc. From Professor Pedreschi’s perspective, AI as a tool for improving the individual leads to greater creativity. However, when applied to a very large community, it promotes conformism and polarization of thought disguised as targeted mass recommendations.
Trend Micro, represented by Camillo Bucciarelli, also spoke, introducing the topic of cybercrime through AI. Camillo outlined two major security issues related to AI:
How can we combat these dangerous new trends? According to Camillo, citing specific examples of Trend Micro products, three methods can be used:
This will be followed by a presentation by Professor Anna Pettini (Professor at UniFi) on a project to improve the assessment of national wealth and implement strategies related to it. The aim of the presentation is to explain that the meaning of wealth decades ago is completely different from today. From a concept of wealth characterized by a sense of possession, we are moving towards a more abstract concept of wealth: mental, economic, and social. The presentation then explored in detail why the assessment of a country’s well-being can no longer be based on GDP calculations, but rather on much more descriptive and truthful factors regarding the well-being of citizens (not just the economy).
There is an indicator that we can roughly describe as a corrected GDP (the GPI, or Genuine Progress Indicator) that helps us understand how positive national growth is for the citizens who should experience its benefits. Professor Pettini’s proposal was to create an AI capable of performing the extremely complex calculations that would provide a view of citizens’ quality of life indicators that is as close to reality as possible.
Among the final speakers at the event, still focusing on personal well-being, was Professor Silvio De Magistris (Professor at UniFi). He spoke about the importance of continuing to focus on non-technical personal skills (what Americans like to call “soft skills”) to maintain a healthy level of well-being in one’s life and relationships, even in the digital age.
The final talk of the event aimed to introduce attendees to a future with physical AI. This form of AI application is already present but often ignored or unaware. The concept of physical AI also encompasses the field of robotics. This means that “physical AI” can be understood as all AI that has a physical aspect as its input or output.
The event lasted all day, and only a few survived the vast amount of information and insights provided by each of the speakers. FlorentIA was an event of great cultural and technological value, from every perspective and for everyone (even those who are afraid of these new applications).
For those who would like to catch the entire event, which was streamed nationwide, the full video is available on YouTube at the following LINK .
S&NT’s promise is to make this event the first of many future editions, with equally exceptional and high-profile guests. This event’s primary purpose was not to publicize itself or its partners, but rather to raise awareness on the issue and initiate action through dedicated working groups.
Nicola Tarlini