Scientists warn AI may erode human judgment rather than trigger a robot apocalypse

Leading AI scholars contend that the most plausible existential threat from artificial intelligence is not a violent takeover but a slow erosion of human mental faculties, termed "enfeeblement." Samuele Vinanzi, a senior lecturer at Sheffield Hallam University, warns that delegating increasingly complex tasks to AI could leave future generations unable to think independently or scrutinize algorithmic decisions. Evidence includes experiments where radiologists trusted fabricated AI recommendations, dramatically lowering diagnostic accuracy, and a near-miss incident where a military AI chatbot generated a false report about a Chinese-flagged vessel.

Critics of dramatic apocalyptic scenarios point out that creating viruses or hacking air-gapped nuclear systems still requires physical human action. Nonetheless, the pervasive use of AI in finance, healthcare, policing, and transportation fuels worries that critical thinking skills will atrophy, especially among students using tools like ChatGPT for assignments. Some technologists, such as Andrea Miotti of ControlAI, caution that if superintelligent AI were to manage core services, loss of human oversight could occur rapidly, turning AI into an autonomous threat.