How AI Systems Impact the Italian Productive Fabric

The topic of Artificial Intelligence is increasingly central in public debate, especially regarding the risks it may pose in terms of job losses. However, it is useful to narrow the discussion to a few more strategic questions. In this sense, the key question we should ask ourselves as a country is the following: where does Italy stand in the implementation of AI within the productive processes of its companies, and what advantages could it gain in terms of real‑economy impact?

Data provides a meaningful picture. AI adoption among Italian companies is growing rapidly, rising from 8.2% in 2024 to 16.4% in 2025 among firms with at least 10 employees, according to the latest Istat figures. This trend reflects an acceleration, though it remains below the EU average of roughly 13.5–18%. In 2025, 46% of professionals surveyed by EY reported using AI tools at work, compared to 12% in 2024, marking a +34% annual increase. Large enterprises (250+ employees) reach 32.5% adoption, while medium‑sized firms (50–99 employees) are at 14%. The main applications involve process automation, data analytics, and predictive maintenance.

Extending the analysis to the patent landscape—although no single, complete official statistic exists for “made in Italy” AI patents—the data from patent filings between 2019 and 2025 reveals a clear pattern: most international and European patent applications by Italian inventors in AI are concentrated in a few regions with high R&D density (Lombardy, Piedmont, Emilia‑Romagna, Lazio, Veneto), with a strong weight of major metropolitan hubs (Milan, Turin, Bologna, Rome).

AI patents fall mainly within classes G06N (knowledge‑based systems, neural networks, machine learning) and G06F (data processing), often combined with application‑specific sectors (healthcare, manufacturing, telecom, finance). Many applications are filed with EPO, WIPO, and USPTO by Italian companies to protect foreign markets; therefore, the “basket” must be examined across these databases, not only at the Italian Patent and Trademark Office (UIBM). Cross‑referencing Italian inventors’ addresses in G06N/G06F classes (2019–2025) with major AI patent families, the national territorial distribution is roughly as follows:

• Lombardy (Milan): first area for international AI patent filings by Italian inventors or companies; strong presence of ICT, fintech, advanced manufacturing, medtech; many patents on image analysis, predictive maintenance, hybrid explainable AI models applied to industry and finance.
• Piedmont (Turin): strong in AI for automotive, robotics, embedded systems, and networks; patents often combine G06N with automotive and telecom classes.
• Emilia‑Romagna (Bologna–Modena): specialization in AI for manufacturing, industrial automation, technical image annotation and analysis, process control.
• Lazio (Rome): driven by major ICT players, banks, and central public administration; patents on explainable AI for financial services, model security, ML pipeline management, hybrid shallow/deep models.
• Veneto (Padua–Verona): AI patents linked to biomedical applications, sensors, industrial IoT, and data analytics for manufacturing SMEs.
• Southern Italy: AI patents exist but are much rarer, often tied to individual university groups or spin‑offs rather than large structured companies.

Looking at AI patent families involving European inventors/companies (a pattern also valid for Italian actors), the typical areas are:

• Computer vision and image analytics: automatic annotation frameworks, active learning on images, models for medical or industrial image analysis.
• Core machine learning and MLOps: multipurpose ML engines, systems orchestrating multiple models, pipelines, and trust/robustness management.
• Explainable AI and control: hybrid shallow/deep systems for explainability, feedback‑loop management between AI systems, robustness evaluation.
• Vertical applications: healthcare (image‑based diagnostics), Industry 4.0 (plant control, maintenance), financial services, telecommunications.

In general, large groups and technology providers focus on core patents (frameworks, ML engines, explainable AI), while Italian manufacturing companies tend to patent vertical applications, often in collaboration with universities or regional research centers.

Impact of AI on Italy’s GDP and Productivity

AI adoption by Italian firms generates a positive impact on GDP through productivity gains, revenue growth, and process efficiency. Estimates indicate a potential annual contribution of up to 8% of GDP in scenarios of widespread adoption. Studies by Censis–Confcooperative forecast +1.8% GDP by 2035 (+€38 billion), while EY notes that 93% of adopting companies report +27% average revenue growth.

Projected scenarios for the impact of AI agents on the national productive system include:

• Conservative scenario: +0.9–1.8% GDP over 10 years (2035), equal to +€38 billion, driven by productivity increases and creation of new AI‑integrated jobs (9 million).
• Optimistic scenario (widespread GenAI): +8% annual GDP at peak within 10 years (+€150–170 billion), stemming from productivity gains (+€120–130 billion), re‑employment (+€40–50 billion), and reduced transition losses (–€10 billion).
• Immediate effects: 52% of top managers report cost reductions and profit increases.

Impact on Employment

The risks of net job losses due to AI introduction are real, but can be mitigated by the creation of new roles within companies. In Italy, 6 million jobs (out of 23 million) could be replaced by 2035, but 9 million jobs could be augmented, and 2–3 million new positions are expected in more technological roles (data scientists, AI trainers). The net impact will depend on reskilling processes and active labor policies.

Reskilling is the critical factor: today, the share of Italian firms using AI is still relatively low (8%). Without investment in reskilling projects, there is a risk of wage stagnation and unemployment rising to 10–15%. With such investments, employment levels should remain stable, with a 2% job growth in the most proactive countries.

Sources: 

• https://www.confapivenezia.it/news/intelligenza-artificiale-cresce-lutilizzo-nelle-aziende-italiane-34-nellultimo-anno/
• https://www.osservatorioai4pa.it/notizie/intelligenza-artificiale-economia-lavoro-giustizia/
• https://www.reviewofailaw.com/Tool/Evidenza/Single/view_html?id_evidenza=2990
• https://www.fortuneita.com/2025/12/21/lefficienza-dellai-e-unillusione-perche-tagliare-i-posti-di-lavoro-penalizzera-le-aziende/