Anthropic Introduces Three AI Development Metrics, Covering AI Involvement in R&D, Agent Oversight, and Compute Allocation

Odaily reports: Anthropic stated on X that as AI system capabilities continue to improve, AI is increasingly participating in the development of next-generation AI systems. To enhance transparency in frontier AI development, Anthropic has released three metrics for tracking AI development: the proportion of AI R&D work completed by AI, the oversight of AI Agents, and the allocation of compute resources.

Anthropic also disclosed interim data on these internal metrics and stated that other frontier AI developers could disclose the same metrics for third-party verification, in order to narrow the information gap between frontier AI labs and the public.