It’s a pleasure to receive and publish the article by university professor Cristina Pronello, who spoke a few days ago at the two-day debate organized in Livorno by Filt-CGIL, an initiative focusing on automation in ports and artificial intelligence.
Artificial Intelligence (AI) is at the centre of the current debate on technological innovation. It is considered as the new revolution of our time, with consumers and companies adopting it much faster than other recent technological innovations like Internet or mobile phones. This process, however, has been going on for years, much longer than we can imagine.
The term ‘artificial intelligence’ was coined in 1956 at the Dartmouth Conference by the American mathematician John McCarthy, together with Harvard mathematician Marvin Minsky and two researchers from the Bell Telephone laboratory, Claude Shannon, and IBM, Nathan Rochester. The 1950s were ‘the dawn of the artificial age,’ in a world dominated by continuous technological breakthroughs and a growing optimism in the power of technology, where ‘the artificial’ was destined to solve nature’s problems, which became ‘the marginalized.’ Then the Cold War boosted investment in AI in the US and, subsequently, other countries joined the AI race that has been happening over the last few years.
Implementing AI, however, calls for a coordinated, socially acceptable approach that cannot be separated from a political responsibility that facilitates the transition of the workforce. If we take a closer look at what emerged at the World Economic Forum, held in Davos in January 2025, two key points emerge in the stakeholder discussion on AI:
1)Technology is advancing faster than the workforce’s ability to acquire new skills;
2)We are still at an early stage, testing and experimenting, because AI is not necessarily going to be adopted in everyday activities. As specialized literature on the subject points out, AI is currently mainly being used for predictive maintenance because it can be based on a sufficient amount of in-house company data. It is also being employed in other predictive activities that regard a number of transport sectors. They often relate to safety and energy management, although most applications are still in the testing phase.
An issue worthy of note regarding the slow adoption of AI is the lack of knowledge of the effects and benefits of this ‘revolution.’ This is because they have yet to be demonstrated and the uncertainties and risks may call its economic sustainability into question. Just how developed technology is, market acceptance and policy changes may influence the cost-benefit ratio.
As an Oliver Wyman survey of 300 global companies shows, 97% of organizations have used AI as a strategic lever for transformation, but only 17% say investment has exceeded expectations. Initial costs are high, retraining and upgrading the workforce is necessary and the transition to new skills for new jobs will take time while the benefits could be very slow in coming. This could challenge investors and policy makers, with inevitable detrimental effects on research.
To give an example, to reap the benefits of AI, workforce strategies are needed, focusing on rapidly evolving skills to change everyone’s mindset, from top to middle management, down to the lowest levels, above all in terms of managing the risks its adoption entails. Indeed, while AI is a technology, its use is an approach to problems and can become a mindset: for example, saving time on certain tasks and reducing personnel costs.
Hence risk management training is linked to teaching people how to use AI, adopting training approaches that show how the key is not being faster in getting information or identifying possible solutions, but in assessing the quality of the information and avoiding over stressing the benefits based on the predictive potential of algorithms, forgetting that these only work well if the data driving them is ‘very big’.
Being overenthusiastic about big data in the transport context is misleading because we forget that we don’t have enough data and that most of it is not digitalized and that we continue to rely on traditional probabilistic models (precisely because there’s only a limited amount of data).
The lack of data is Artificial Intelligence’s Achilles’ heel. This has fuelled a data market in the hands of a few big companies that have a monopoly on it. Global players like Google, Apple, Microsoft, Amazon, Meta – just to name the most important ones – are the ones who collect personal data and profile people through their services.
The operating systems (Android and IoS) do not allow other companies to track people or collect the same data that they gather, and this produces a twofold effect: monopolizing data through eliminating competition and making it impossible for other companies to innovate themselves through data. The fact that Google and Apple developed the operating systems of the devices we use gives them a competitive advantage that they have no intention of losing, because data is like oil these days, a source of immense profits.
At a time when we speak of open markets and competition, we are faced with a paradox that only politics could interrupt, with data tracing measures, data sales and the proper taxation of companies from other countries that make large profits in Europe. The current European privacy policy, paradoxically, penalizes European companies in favour of non-European ones, with the consequence of not protecting (if not virtually) the privacy of citizens and not guaranteeing cybersecurity. It is not far-fetched to think that people might come to realize that they are a source of great profit without taking real advantage of it and generate a shift that could penalize digital in favour of analogue, making data gathering through digital devices ineffective.
Something else we should think about is that adopting artificial intelligence not only differs from sector to sector, but especially from country to country, with a large gap between Europe and non-EU countries, between the United States and Asian countries. Oliver Wyman’s Forum AI index, published in January 2025, shows how the AI Index identifies eight countries as AI leaders, with the US and China respectively in first and second place. European countries show varying rates of adoption, with only France and Germany in the top eight, but at the bottom of the list. It is clear how uneven the rate of adoption of AI is around the world, contributing to exacerbating differences and increasing tensions. In fact, while most of the literature on Artificial Intelligence focuses mainly on technological aspects, it has multiple facets ranging from culture, society, workforce, environment, politics and geopolitics.
The AI race is mainly driven by economic reasons and the quest for supremacy. This is the biggest risk, especially for its geopolitical and ethical implications: speculation risks and overemphasis on data science, neglecting critical thinking and overestimating trust in algorithms. Without carefully designed training programs that take into account the multifaceted nature of this technology, we risk failing to adequately prepare this generation and the ones to come for the future.
Before implementing AI, there has to be a major effort at various educational levels (from primary to university) and then at all levels of the workforce: preparing people to think independently, to be creative and to be social human beings (collaboration instead of competition).
Finally, in the transport industry, it is essential to improve the quality of services before applying AI because, otherwise, the greatest risk is to neglect the basic needs of the sector in favour of a technological illusion. Taking funding away from basic services to invest in AI means neglecting user needs, not taking care of them … and users are transport systems’ customers.
Translation by Giles Foster
