Inteligencia Artificial (IA)
Pau Vila: "AI excels in training memory, but fails in real-time conversations" - ON ECONOMIA
Gianro Compagno
2026-04-04
5 min read
In a recent conversation with ON-IA, Pau Vila, CEO of LC Paper and collaborator of ON Economia, analyzes the role and scope of artificial intelligence in the business environment. Vila, also president of the Institut Ostrom and vice president of FemCAT, emphasizes that AI represents an evolution of machine learning, capable of providing varied responses through probabilistic calculations, although without a real understanding of context. According to Vila, the true value of AI lies not so much in conversational interaction, like that popularized by ChatGPT, but in the tangible results it produces: from software to images and tools that can revolutionize internal processes of companies.
Vila highlights recent advancements, such as Anthropic's Claude model, which excels particularly in software generation. He also warns about the speed and inequality with which companies are adopting AI, noting that failing to adapt could lead to a loss of competitiveness. Despite the challenges, he argues that AI opens significant opportunities to optimize processes and develop customized solutions at much lower costs and timelines than traditional methods.
For Vila, artificial intelligence is not a new concept, but an extension of disciplines like machine learning, which have been studied for years. He explains that the key difference is that current AI not only reproduces what it has learned but can infer and offer multiple valid solutions, depending on its training in specific areas.
Regarding the impact on LC Paper, Vila acknowledges that his perception has evolved over time. Initially, the focus was on conversational interaction, but now AI is used to generate products and services that transform business operations. The focus, he asserts, should be on the results that AI can deliver, beyond simple conversation.
On specific cases, Vila highlights the emergence of Anthropic and its Claude model, which has surpassed ChatGPT in downloads and specializes in software creation. He explains that, although Claude does not generate images or videos, its ability to develop programs is remarkable, allowing for solutions that previously required large investments and long development timelines.
Vila also shares his personal experience in creating software tools for business management, always prioritizing those solutions that, with a reasonable investment, bring about significant change.
From his perspective at FemCAT and the Institut Ostrom, Vila expresses concern about the inequality in AI adoption within the business fabric. He advises companies not to settle for past experiences and to explore the current possibilities of AI again, as competitiveness can be quickly lost if one does not keep pace with the market.
Finally, he warns that AI encompasses a very broad spectrum and can generate both risks and benefits. He underscores the importance of working on technical and security aspects to minimize problems and maximize the opportunities that this technology offers.
Source: elnacional.cat