Harvard psychologist Steven Pinker argues that the existential risks posed by artificial intelligence are currently overstated. In an open letter to blogger Scott Alexander, Pinker dismissed the 'paperclip maximizer' scenario, noting that such theories incorrectly conflate intelligence with a desire for domination.
While Pinker maintains an optimistic view on AI, he admits to underestimating the capabilities of large language models and the reckless pace at which companies release new products. He noted that many of those driving this competitive release cycle are the same individuals who frequently warn about potential AI-related dangers.
Pinker identifies bioterrorism, cyberattacks, and unregulated AI agents as the primary risks that require attention. He advocates for rigorous safety engineering, emphasizing the need for independent oversight, clear accountability mechanisms, and strict human control over AI systems.
Scott Alexander, who previously challenged Pinker to a debate, holds a different perspective. In June, he estimated the probability of human extinction caused by AI at 20 percent. Alexander supports increased investment in safety research and suggests slowing down AI development through international agreements.
