Artificial intelligence, companies try to catch up

From manufacturing to life sciences. With the main business functions involved: from human resources to customer service. In a collection of 241 use cases applied in 76 companies, Confindustria with the Artificial Intelligence Sounding board led by special advisor Alberto Tripi focuses on the first benefits that industry is capitalising on from the use of AI.

In Italy, the adoption of artificial intelligence is going through a phase of appreciable but still limited progress characterised by a certain lack of homogeneity across sectors, territories and size classes. According to Istat, in 2021 only 6.2% of Italian companies with at least 10 employees used AI. The share dropped to 5% in 2023, but rose again to 8.2% in 2024. There is a recovery then. However, we are still far from the EU average (13.5%) and the lag is most marked in SMEs, where only 1.4% use AI in at least three business functions. Adoption is still fragmented: strong in the telecommunications (27.6%) and electronic equipment (15.7%) sectors, but much weaker in textiles and fashion (4.6%).The main barriers remain high costs and digital skills shortages.

Having said this, the compendium of concrete applications and impacts measured by Confindustria offers in conclusion five general indications, valid for all sectors companies must equip themselves with representative and accurate datasets in order to effectively train the algorithms and obtain reliable results; all personnel must be adequately involved and informed from the earliest stages of the project; digitisation platforms and IoT (internet of things) systems functional to AI applications must be adopted promptly; a gradual and pragmatic approach is preferable, starting with limited pilot projects; in any case, human oversight of the solutions introduced must be ensured. While these five principles are transversal, each sector presents explicit peculiarities in the use cases examined (21.6 per cent refer to health and life sciences; 20.7 per cent to manufacturing; 17.4 per cent to sustainable mobility; 6.6 per cent to the public sector; 5.4 per cent to tourism; 28.2 per cent to other sectors or are multi-sectoral in nature). In total, the cases using generative AI were only 18% of the total, demonstrating that in the corporate field even the lesser-known applications of ChatGPT and its epigones have a concrete relevance.

In the field of health, the applications introduced include the automation of clinical documentation, diagnostic imaging, clinical decision support systems, telemedicine and remote patient monitoring, preventive and personalised medicine, and the analysis of biological networks to identify the most appropriate therapies for each patient. The manufacturing sector makes extensive use of digital twins for predictive plant maintenance and warehouse management through computer vision systems, which is also used to expand the capabilities of robots used in factories. At the same time, AI can predict the production costs of new products through the analysis of historical databases. In the field of sustainable mobility, the virtual replication of the physical infrastructure (the digital twins mentioned above) enables real-time monitoring of performance by optimising energy consumption, but other applications concern the optimisation of the 'mileage curve' of vehicles and the prevention of the risk of component failure in buses and on-board systems. For public administration, the use of AI has so far meant 'smart city' efficiency, waste management, road and lighting maintenance, and control of large natural spaces. In tourism, these range from dynamic price optimisation to the management of flows during peak periods, from the optimisation of public services linked to the stay to generative AI for reviews and itineraries. In customer service, the most consolidated uses concern advanced chatbots for assistance, analysis of user opinions, automatic ticketing.

Analysing business functions instead, AI is proving most effective in the areas of research & development, administration, marketing and sales. Weaker, on the other hand, is the adoption in production processes, logistics and ICT security.