The article at a glance

  • Generative AI is growing rapidly, accounting for 43% of the artificial intelligence market in Italy, although adoption among SMEs remains limited.
  • SMEs face technological, economic and organizational barriers that slow down AI integration compared to larger enterprises.
  • There are risks related to misinformation, including unrealistic expectations, improper use and lack of supervision, which can lead to poor decisions.
  • SMEs can benefit from AI, provided it is adopted with a conscious and supervised approach.

The AI market in Italy: where we stand

In recent years, the artificial intelligence (AI) sector in Italy has experienced significant growth, reaching a value of €1.2 billion in 2024, with a 58% increase compared to the previous year (Artificial Intelligence Observatory, POLIMI School of Management).

A significant share of this growth is driven by Generative AI (GenAI), which accounts for 43% of the market and is mainly based on two types of algorithms: large language models (LLMs) for text generation and algorithms for image creation.

However, while large enterprises have already embraced AI through structured initiatives, small and medium-sized enterprises (SMEs) are still at an early stage.

The data is clear: 59% of large companies have launched AI projects, while the percentage is much lower among SMEs, despite more than half of them showing interest in the topic.

Specifically, only 7% of small enterprises and 15% of medium-sized enterprises have started initiatives related to artificial intelligence.

Why do SMEs adopt AI more slowly than large enterprises?

Only 8% of Italian SMEs have adopted Generative AI tools available on the market through licensed solutions. These companies are largely those already active in the AI sector, while a smaller share consists of organizations that are beginning to experiment with the technology with very limited investments.

What explains the gap in adoption between large enterprises and SMEs? We have identified three main factors:

1. Technological maturity

Large enterprises can build experience in using AI more quickly, thanks to greater availability of resources. SMEs are only starting to explore its potential.

2. Project complexity

Initiatives in large organizations rarely involve AI alone. They often integrate multiple technologies, making projects more complex. This level of complexity requires highly specialized skills, which are harder to find within SMEs. Large enterprises typically rely on management teams with specific expertise, while SMEs depend on more versatile managerial roles that are often less focused on specialized competencies.

3. Economic and decision-making constraints

SMEs are significantly more sensitive to cost. For example, a large enterprise can invest in strategic consulting projects worth hundreds of thousands of euros, while SMEs often need to rely on more accessible solutions, sometimes at the risk of choosing inadequate tools.

AI misconceptions: the risk of misinformation

SMEs are currently exposed to misleading narratives about what AI can actually deliver. In some cases, communication has created excessive enthusiasm, leading to unrealistic expectations or, conversely, to unfounded concerns.

One of the most widespread misconceptions concerns the diagnostic capabilities of tools such as ChatGPT and other language models, which are sometimes presented as superior to doctors in diagnosing diseases. In reality, these tools do not have clinical capabilities and cannot replace professionals. They can only support decision-making.

Another example of improper AI use can be found in some U.S. courts, where automated drafting of legal decisions has been tested, raising questions about fairness and accountability in decision-making processes. A further case involved the airline Air Canada, which, after a legal dispute, had to compensate customers due to incorrect information provided by an unsupervised chatbot in its customer care service.

Even more concerning is the use of AI in therapeutic contexts, where some chatbots have been suggested as support tools for patients with psychological disorders, leading to serious consequences. These examples highlight the risks of assigning critical responsibilities to systems that lack empathy and autonomous judgment.

How can SMEs adopt AI effectively?

While the risks associated with AI must be considered from the outset, there are many scenarios in which Generative AI can deliver real value to SMEs. Some examples include:

  • AI can support the reading and structuring of complex texts, improving decision-making processes and enabling more effective analysis and management of large volumes of data.
  • SMEs can automate repetitive tasks using LLMs to organize information, generate summaries or support customer care activities, reducing operational time.
  • AI can be used to analyze large volumes of customer data, helping companies identify relevant segments, predict purchasing behavior and optimize marketing strategies.

To distinguish between safe and risky use of AI, SMEs can follow two simple recommendations:

  • Avoid relying on AI for opinions or internal information, as generative models may produce unverifiable content.

  • Do not allow AI to make critical decisions autonomously. All generated content should be reviewed by a human before being used in business contexts.

Like any tool, the value of Generative AI depends on how it is used. SMEs can benefit from it by adopting a conscious approach and avoiding misleading expectations. Human supervision remains essential: AI can support decisions, but it cannot replace them.