New Research Highlights Limitations in AI Agents' Independent Research Capabilities

A recent paper indicates that while advanced AI agents are proficient in coding research, they currently demonstrate limitations in their ability to autonomously conduct research. This finding suggests a present constraint on the independent scientific inquiry capabilities of frontier AI systems, particularly in complex and open-ended research tasks.