A new study published in ScienceDaily reveals that even the most advanced AI models fail the classic Stroop test as word lists grow longer. This research highlights significant cognitive bottlenecks in artificial intelligence compared to human attention and processing. The findings suggest that while AI can simulate complex reasoning, it lacks the fundamental inhibitory control mechanisms that humans use to manage conflicting information. Source: https://www.sciencedaily.com/releases/2026/06/260610003049.htm
Psychiatrists Warn of Generative AI's Double-Edged Role in Mental Health Clinics
A cross-sectional survey of 408 psychiatrists published in the Journal of Medical Internet Research spotlights the dual nature of generative artificial intelligence in mental health settings. While AI can serve as a fatigue-free support tool that helps patients communicate complex emotions, unsupervised chatbot interaction threatens to reinforce patient delusions, ignore non-verbal cues, and trigger crisis situations without a safety net. Researchers underline the immediate need for robust clinical validation, governance, and professional supervision before wider adoption.