AI Scientists Now Design Experiments and Write Papers

Artificial intelligence has crossed a remarkable threshold: it can now conduct scientific research almost entirely on its own.

Three landmark AI systems reported in Nature in 2026 show just how far machine learning has come. The Empirical Research Assistant can develop scientific software for everything from cell analysis to disease forecasting. The AI Scientist generates research ideas, writes code, conducts experiments, creates figures, drafts manuscripts, and even performs peer review. MIRA works in medicine, interacting with simulated health records to order tests, diagnose patients, prescribe treatments, and recommend hospital admissions.

Professor Guang-Guo Ying of South China Normal University examined these systems in a new article published in Artificial Intelligence & Environment. He notes that AI has evolved from a passive tool into an active participant in scientific discovery.

The systems aren't perfect. They can generate fake citations, write incorrect code, duplicate figures, and produce results that are physically impossible or difficult to reproduce. These flaws highlight why human oversight remains essential.

The Ripple Effect

This technology could dramatically expand humanity's ability to explore new ideas and test hypotheses at unprecedented speed. Imagine thousands of research questions being investigated simultaneously, freeing scientists from repetitive tasks to focus on the creative work that machines can't do.

Ying envisions a future division of labor where humans define important questions, establish ethical boundaries, evaluate ambiguous findings, and decide which ideas deserve pursuit. AI handles the computationally intensive and repetitive exploration that would take human researchers years to complete.

The goal isn't to replace scientists but to create a partnership. Human curiosity and judgment combined with machine scalability could accelerate solutions to pressing challenges in medicine, climate science, and technology.

The scientific community now faces an important responsibility: guiding this technology's development toward openness, safety, and responsible discovery. How researchers choose to integrate AI into their work will shape the future of scientific progress for generations.

The age of the AI scientist has begun, and its promise depends on keeping human wisdom at the center.