Recently, OpenAI unveiled two comprehensive articles. One article announced that the company had successfully implemented 'automated AI research interns' prior to September of this year, thereby realizing Altman's objective set the previous year. Furthermore, OpenAI outlined its ambitious plan to develop an 'automated AI researcher' by March 2028. According to internal data, a single workday for a human member of the R&D team is equivalent to 3.1 days of work accomplished by an AI Agent. However, this efficiency comes at a cost: the average daily inference expense for an AI Agent exceeds $600 for ordinary researchers and surpasses $7,000 for the top 10% of researchers. At present, AI interns are adept at managing labor-intensive tasks, yet strategic decision-making still necessitates human oversight, as evidenced by the mere 20% success rate of long-term tasks without human intervention.

The second article, penned by Chief Scientist Jakub Pachocki, asserted that machine intelligence has now eclipsed human intelligence, urging immediate global action in response. Adding to the discourse, Jensen Huang declared that Artificial General Intelligence (AGI) has indeed arrived and pledged to supply OpenAI with 400,000 GPUs. OpenAI has already harnessed over 100,000 GPUs in the development of GPT-6 Astra, marking a significant milestone in the realm of AI. The trajectory is clear: the process of AI creating AI is now an unstoppable force.