Nature: Human Scientists Still Trounce Best AI Agents on Complex Research Tasks — Study Quantifies the Gap

Nature·medium signal

A study published in Nature finds that human scientists significantly outperform the best AI agents on complex research tasks, providing empirical evidence for the gap between AI coding/reasoning benchmarks and actual scientific research capability. The finding is relevant context for the current wave of autonomous agent hype and suggests that agent orchestration and human-in-the-loop designs remain essential for serious research workflows.