Leading for two years but stubbornly covering up: OpenAI product leader reveals insider information about job hopping, bluntly stating that his former employer DeepMind is slow to take action According to Beating monitoring, Thibault Sottiaux, the head of OpenAI's core products, revealed in an interview with Wired that he had jumped from Google DeepMind to OpenAI in anger. Sottiaux revealed that before ChatGPT was released at the end of 2022 and ignited the global wave of generative AI, DeepMind had already mastered and "boxed" similar big language models and dialogue technologies for two years internally, but the conservatism and bureaucracy of the top management prevented the technology from being launched. Sottiaux said that the emergence of ChatGPT in 2022 completely woke him up and made him realize that he could no longer watch good technology being "spoiled" in the hands of his former employer. He admitted, 'This (big language modeling technology) is something we have been sitting on at DeepMind for almost two years, but we just didn't take action at that time.'. This' inaction 'made him determined to move to San Francisco and join OpenAI to implement the technology. Prior to this, he worked at Google's London office and later joined DeepMind to oversee the development of research infrastructure for projects such as AlphaGo. Now, Sottiaux has been promoted to the head of OpenAI's core products, reporting directly to Greg Brockman, and fully leading the major redesign of ChatGPT and its integration with the Codex intelligent agent sandbox. His revelation once again confirms the industry's widespread suspicion that Google is not technologically backward, but has forcefully given the lead to faster moving OpenAI due to slow internal decision-making, concerns about the erosion of search fundamentals, and excessive prevention of security risks. [Original link]