Google DeepMindのCo-Scientist、本格的な研究パートナーへ進化

google deepmind has upgraded its multi-agent system co-scientist from a hypothesis generator to a lab-integrated research partner. co-scientist now plans experiments, writes code, and controls lab equipment, delivering experimentally validated results across three disciplines: materials science, biology, and computer science. in initial tests, co-scientist designed synthesis recipes for materials science, built predictive pipelines in biology with expert feedback, and independently developed a medical ai architecture that outperformed gpt-5 in benchmarks. the system uses verification modules to cross-check numerical claims against execution logs, reducing fabricated results. co-scientist operates in three phases: ideation, experimentation, and paper generation.