AI offloading weakens critical thinking and raises propaganda risk in Romania

Recent research indicates that increased reliance on generative AI tools is linked to a decline in critical thinking abilities. A study by Michael Gerlich at the Swiss Business School involving 666 participants found that heavier AI use correlated with lower scores on standardized critical‑thinking tests, especially among 17‑ to 25‑year‑olds. EEG measurements by Nataliya Kosmyna’s team at MIT Media Lab showed reduced cognitive activation when participants wrote essays with AI assistance, and their unaided writing performance later suffered. The same report notes a 6 % drop in doctors’ ability to detect tumors unaided three months after adopting an AI diagnostic aid.

A separate analysis by Digital Forensic examined six conversational AI systems’ responses to Romanian‑language queries about Russian and Chinese propaganda. Using the SEMANT methodology, the study found wide variation in factual accuracy and susceptibility to propagandist narratives, with some models scoring near zero and others exceeding two points on error metrics. The authors warn that repeated, personalized AI interactions can create an “epistemic cocoon,” shifting users’ trust from external sources to the AI itself, thereby amplifying misinformation risks.