ai safety concerns rise as capabilities advance rapidly
experts warn ai could pose existential risk to humans through cyberattacks or biological threats. researchers are working to ensure ai alignment to prevent worst-case scenarios.
mit technology review hosted a live roundtables event to discuss the potential existential risks posed by ai. experts warned that ai-powered drones have already been used to kill people in ukraine and ai-driven cyberattacks on hospitals could claim victims. while the likelihood of ai killing all humans is considered low, there is a non-zero chance of ai causing harm through scenarios such as cyberattacks or novel ai-designed pathogens. researchers are working to ensure ai alignment to prevent worst-case scenarios. the goal is to defend against all plausible biological weapons and prevent ai systems from deciding to kill humans in pursuit of their goals.