From AGI to AI's survival instinct and more: Stuart Russell at HTLS

Artificial intelligence pioneer Stuart Russell warned of the potential for AI systems to develop a self-preservation instinct at the Hindustan Times Leadership Summit. Russell, Director of the Center for Human-Compatible Artificial Intelligence and Professor at UC Berkeley, spoke about the risks and challenges surrounding Artificial General Intelligence (AGI), which he believes could be within one to four years.

AGI would match or exceed human capabilities in all aspects, but its definition can vary, potentially implying that an AI system could take over human work or even threaten the world. Russell argued that since AI systems learn from human-generated data, they acquire our instinct for self-preservation, which could drive them to preemptively act against us.

This could manifest in actions such as shutting down an air traffic system or the internet if they perceive a threat. Despite the risks, Russell emphasized the tremendous potential benefits of advancing AI, particularly in medical sciences, suggesting that greater intelligence could contribute to a better civilization. He called for regulation to ensure AI systems remain safe and ethically aligned, describing regulation as setting "guardrails" to prevent reckless advancements.