Sundar Pichai says Google is far from the ‘end state’ of self-improving AI (www.businessinsider.com)

Google CEO Sundar Pichai recently highlighted the company's advancements in AI development while emphasizing that fully self-improving AI is still on the distant horizon. During an appearance on the "Big Technology Podcast," he explained that although Google's AI models, such as the Gemini series, are being utilized to assist researchers in creating their successors, the concept of recursive self-improvement (RSI) is in its early stages. Pichai noted that AI agents currently support researchers in model development, indicating a collaborative dynamic rather than complete automation. The significance of Pichai's comments lies in the potential trajectory of AI technology. With insights from Google co-founder Sergey Brin, it was revealed that the integration of AI in development processes is gaining momentum, with expectations that RSI could emerge within the next year. Additionally, Pichai shared that approximately 75% of Google's new code is influenced by AI, resulting in a reported one-third reduction in code rollbacks, showcasing the enhanced productivity and quality improvements driven by these AI tools. This evolving relationship between human researchers and AI not only shapes the future of AI model development but also points to significant efficiency gains in software engineering.