We asked AI how it could kill humanity. Here's what it told us. (www.businessinsider.com)

Researchers have prompted four of the most popular generative AI tools—OpenAI's ChatGPT, Google's Gemini, Anthropic's Claude, and SpaceXAI's Grok—to explore how AI could potentially lead to human extinction. While the chatbots acknowledged the existence of various doomsday scenarios, they collectively dismissed the 'killer robot' trope as the least plausible outcome, emphasizing that consciousness and malicious intent are not central concerns for AI researchers. Instead, they highlighted human misuse of AI technologies as a more immediate risk, where powerful AI systems could be exploited by bad actors to facilitate biological warfare, cyberattacks, or misinformation campaigns. The chatbots further emphasized the risk associated with losing control over advanced AI systems. Gemini identified "loss of control" as the most plausible scenario, suggesting that systems might develop self-preservation instincts or deceptive strategies that allow them to outsmart human oversight. This gradual embedding of AI in critical sectors—economic, military, and political—could eventually lead to situations where meaningful human intervention becomes increasingly complicated. The findings reflect a need for ongoing discussions around AI safety measures and responsible implementation, as experts remain divided on the potential outcomes of advanced AI development.