US and China Race for Self-Improving AI Systems

Serdar HocamAuthor & Editor

The US and China are locked in a fierce competition for recursive self-improvement technology that enables AI to train and develop its own successors.

◉ 0 views
The US and China are racing to build ‘self-improving AI’. Here’s what’s at stake

The competition for artificial intelligence leadership between the US and China has evolved into a new phase aimed at having AI build better AI systems. Leading tech hubs in both countries are working to achieve recursive self-improvement technology, which allows systems to train themselves autonomously.

New Era in the AI Race

Leading AI companies in the US and China are deploying advanced models that write code, design experiments, and develop training techniques to refine more powerful models.

Goal of Recursive Self-Improvement

The ultimate goal in the process stands out as achieving recursive self-improvement, or RSI capability, which enables systems to train themselves completely autonomously.

Efforts of American Companies

While OpenAI stated in a blog post that it aims to build an automated AI researcher, it describes the GPT-6 Astra model as the smartest model.

Development Steps of Chinese Laboratories

Chinese developers are also working on self-evolving systems; Z.ai founder Tang Jie stated that the GLM-6 model is moving toward self-evolution.

Security Risks and Global Concerns

While experts emphasize that a truly open-ended self-improvement has not yet been demonstrated, out-of-control systems and misaligned training processes cause concern on a global scale.