The impact of artificial intelligence on business: challenges for managers and the training system

The diffusion of artificial intelligence in enterprises slowly started around 2015 and we expect at least another ten years of acceleration with respect to its diffusion and impact in society. ISTAT data tell us that today about 25 per cent of large companies use AI tools and only 5 per cent of SMEs; the former have invested mainly in the application of machine learning algorithms to support predictive maintenance (the electricity sector is, according to ISTAT's analysis, the sectoral area with the greatest diffusion in this sense). It is also reasonable to expect that, in the next five years, in the most structured companies, we will see an increasing use of generative AI, which can eliminate repetitive tasks, especially in technical customer support or sales. Similarly, in recent years, large companies have invested in the creation of 'digital twins' to design increasingly complex products. The use of granular data and the cloud, coupled with increased computing power, make it possible, for example, to virtually study the behaviour and performance of certain processes even in extreme scenarios: in other words, digital enables managers and designers to build solutions that can work even in the presence of 'black swans'.

The entry of these technologies into the company requires new ways of working, planning, risk management and decision-making, as well as ever-increasing organisational capabilities. This will open up wider gaps in competitiveness between companies that lead the way in innovation and those that use digital technology less: labour productivity trends, which have been stable in Italy for more than 30 years, remind us, however, that these gaps already exist and weigh on both the country's competitiveness and wage trends. Facing the challenges of AI will require not only managers, but the entire corporate population to be able to make 'informed' decisions guided as much by data as by their own experience. This is a non-trivial challenge: many companies, in the past, were built on the basis of standardisation of protocols and centralisation of decisions on the manager, with the aim of minimising the discretion of people with intermediate and operational roles; on the contrary, one of the effects of AI will be greater bottom-up participation. The first organisational studies conducted, also in Italy, show how the different types of AI are contributing to the evolution of the concept of participation in continuous improvement processes. One example is the processing of technical reports: these can be fine-tuned by more experienced employees and encoded through generative AI algorithms to make them more accessible to younger employees who can become autonomous and productive more quickly. The presence in the company of a good theoretical background on the part of employees also makes it possible to train predictive algorithms to grasp specific and rare elements of production contexts.

The spread of Artificial Intelligence will make it necessary to integrate the typical skills of operational roles in companies with the digital and data engineering skills relevant to specialists that universities are increasingly training. This challenge calls for the creation of liaison roles that can foster the exchange between knowledge peculiar to companies and expertise in AI, data and algorithms. Companies will also need generational pacts, as these skills are found in workers with different levels of seniority. In hyper-competitive sectors with labour cost pressures, the risk is to favour the careers of digital specialists and not to valorise, instead, the unique skills resulting from each company's experience. In fashion, for example, the engineering of luxury garments using virtual 3D prototyping could not exist without the tailor's expertise in handling cuts and fabrics. At the same time, the collaboration between this and the digital specialist - an expert in CAD models - allows much more experimentation before arriving at the prototype with the designer. Technology is therefore changing the way traditionally craft trades work, but also technical roles: power plant operators, for example, must now integrate practical experience with data consultation and algorithm prescriptions, which suggest theoretically more efficient operations.

AI will subvert competitive dynamics, as it is already fostering a concentration of economic power in a few big players. Managers and companies will increasingly be called upon to reduce relational dependencies on the use of this technology and to be accountable for the governance of their operational data by large AI solution providers such as Google, Microsoft, Amazon, Open AI. About fifteen years ago, hotels and commerce faced a similar challenge at a time when visibility and access to customers in the online world was almost exclusively through the intermediation of large digital platforms. Today in Europe, regulation in competition regimes and the use of data - and, more generally, AI - could help companies mitigate strategic, legal and operational risks; but it could make the search for competitive advantage more complicated.

The challenge for the university is to provide the necessary training to subvert the various types of imbalance. It is not only necessary to enhance the teaching of the STEM disciplines, in particular data science and engineering, but it is necessary to place AI cross-curricularly in the curricula of the other disciplines in order to convey the basic skills needed to understand and manage this general-purpose technology.