Italian AI expert Walter Quattrociocchi warns against overstated robot consciousness

Walter Quattrociocchi, director of the Center for Data Science and Complexity for Society at Sapienza University in Rome, has been appointed by the European Commission as one of sixty experts monitoring the EU AI regulatory framework. He argues for a "healthy skepticism" toward claims that AI systems possess consciousness, coining the term "epistemia" to describe the mistaken belief that plausible language generation equals knowledge. Quattrociocchi stresses that large language models (LLMs) lack metacognition and cannot self‑assess the reliability of their outputs, citing the failed Italian AI project Emma‑5 as an illustration of these limits.

In parallel, Nobel laureate Giorgio Parisi’s recent collaboration with the AI model Claude demonstrates how artificial intelligence can serve as an instructional partner rather than an autonomous master, helping resolve a decade‑old problem in the theory of jamming. The partnership underscores that human expertise remains essential for verifying AI‑generated results.

Cyber‑security specialist Bernardino Grignaffini of the Certego unit (now part of the VEM Group) warns that the same AI technologies give attackers a decisive edge. According to his PanOptikon platform, Italian firms monitored in 2026 faced 7,552 security incidents and over 58,300 alerts across 1.2 million digital assets, illustrating a shift toward AI‑driven automation of phishing, vulnerability scanning and adaptive malware. Grignaffini calls for unified, real‑time defense models to close the emerging gap.