Claude did best on a new benchmark for agents that build agents. It still passed fewer than a quarter of the tests.

The benchmark results for 'agents that build agents' show that even the top-performing AI, Anthropic's Claude, only passed a small fraction of tests. This indicates that while progress is being made in AI's ability to autonomously develop other AI systems, the technology is still in its early stages. For Asian tech companies, particularly those investing heavily in AI research and development, this suggests that fully autonomous AI agent creation is still a distant goal, requiring continued significant investment in foundational AI research rather than immediate application. The observed limitations mean that human oversight and intervention will remain crucial in AI development pipelines for the foreseeable future, impacting resource allocation and skill requirements across the region. The focus should remain on incremental advancements and robust testing, as the current state of the art still has substantial room for improvement.