Google CEO Sundar Pichai: Full Recursive AI Self-Improvement Remains Far Off

IT Home reported on October 24 that, according to Business Insider, Google CEO Sundar Pichai believes there is still a long way to go before AI can fully and autonomously improve itself.

Pichai revealed that Google researchers have begun using AI agents to assist in developing models, but the technology remains in its infancy. "We are far from the end state of recursive self-improvement (RSI). To be precise, across various stages of model development, AI researchers have already begun working collaboratively with self-improvement loops. You could say AI agents are assisting researchers with a certain degree of automation to co-develop models." Pichai expects to see further progress next year.

Google co-founder Sergey Brin mentioned earlier this month that the company's Flash series models and Gemini 4 have already demonstrated AI's ability to participate in its own development. Currently, researchers are working alongside AI models to develop the next generation of models.

According to the report, IT Home learned that in April this year, Google disclosed that 75% of its new internal code was already generated by AI. In Pichai's view, however, simply tracking the proportion of AI-generated code no longer accurately reflects the reality. "To us, that metric doesn't mean much anymore, because now every piece of code we touch is deeply influenced by AI."

Pichai also noted that Gemini 4 Argon will help people tackle more complex programming challenges. Additionally, the frequency with which Google has had to roll back software to older versions due to bugs or system failures has dropped by a third. "Therefore, I believe AI tools are not only boosting programming efficiency, but also improving code quality."