Anthropic's AI Model Claude Participates in Developing Its Own Successor

AI lab Anthropic has announced that the Claude model is helping develop the next, smarter version, leading 26% of model R&D under human supervision.

Claude's Role in the R&D Process

According to the statement made by Anthropic, Claude leads 26% of the company's model research and development activities.

Starting from a high-level prompt, the model can complete the majority of tasks end-to-end under human supervision.

Human Supervision and Collaboration

Approximately 90% of the company's research and development activities are carried out in collaboration with Claude.

The model can handle large workloads under strict human guidance, but it does not yet operate completely autonomously.

AI Safety and Transparency

Industry leaders, notably Anthropic CEO Dario Amodei, are calling for a slowdown in the development of the technology due to safety concerns.

The company argues that the gap between what leading labs know and what the public knows must be minimized.

Recursive Self-Improvement

It is noted that models accelerating their own development could make it harder for humans to understand or control these systems.

These shared metrics can help understand how close leading AI labs are to recursive self-improvement.

Rapidly Increasing Contribution Rate

In February, the rate at which Claude led research and development work was at zero.

Six months later, as of August, the model has reached a position where it leads a quarter of these efforts.

Supervision Measures and Debates

It was reported that approximately 30,000 agents were deployed in research and engineering work as of August.

While safety debates continue in the industry, some leaders support slowing down development, while others oppose it.