Google AI leader sees new path to superintelligence through scaling compute

A Google AI leader, Logan Kilpatrick, believes that scaling test-time compute could lead directly to artificial superintelligence (ASI). He stated that this approach appears increasingly likely, contrasting with previous views that focused on gradual milestones. Kilpatrick noted that achieving artificial general intelligence (AGI) may resemble a product release rather than a sudden breakthrough. He mentioned that both Google and OpenAI are exploring new methods to enhance AI reasoning abilities. Kilpatrick's perspective has shifted, now considering the direct approach advocated by OpenAI cofounder Ilya Sutskever as potentially effective. He previously led developer relations at OpenAI and joined Google earlier this year.