Summary
A recent publication in Nature discusses an agentic system that successfully developed concepts from two computer-science papers, though the original authors were reportedly unimpressed. The article suggests that AI is not yet ready to research itself.