Recently, Lilian Weng, the co-founder of the AI startup Thinking Machines, declared her resignation, citing health concerns as the primary reason. She had been on a month-long leave of absence due to illness. In her resignation statement, Weng revealed that she had experienced frequent bouts of sickness over the past seven months, rendering her physically unable to cope with the relentless high pressure and demanding workload associated with a startup environment. However, a mere two days after her resignation announcement, Weng confirmed her return to OpenAI. There, she will take charge of the Recursive Self-Improvement (RSI) team, spearheading efforts to enable AI to autonomously design, train, and evaluate next-generation models. Weng's rejoining is in perfect harmony with OpenAI's strategic vision of crafting a fully automated "AI researcher." This vision aims to achieve model self-evolution through a bottom-up approach of algorithm iteration and computational optimization during the testing phase. Before her stint at Thinking Machines, Weng had an illustrious seven-year career at OpenAI. During this period, she led applied AI research and oversaw the development of safety systems teams. Additionally, she actively contributed to safety initiatives for models such as GPT-4. In her new role upon returning, Weng will focus on constructing a technical pipeline for the "automatic evolution and mass production of GPTs," rather than directly engaging in the development of the next generation of GPT models.