OpenAI放弃全面自动化：人类仍是关键
openai現在預計到2028年ai將與人類研究員“協同”工作，而不是實現全面自動化，這標誌著其agi路線圖的重大轉變。
openai reversed its previous target to build a fully autonomous ai researcher by 2028, now claiming that 'a significant fraction' of research will involve humans working 'in tandem' with ai. altman and pachocki argue that full automation would be dangerous and unfulfilling, emphasizing that humans must remain responsible for deciding what is worth doing. simultaneously, the company proposed an international organization to coordinate ai efforts and enforce slowdowns of frontier development if societal resilience cannot catch up. this pivot suggests openai is pivoting from pure autonomy to becoming an implementation partner, likely due to the technical difficulty of deployment and slower-than-expected adoption.