Autonomous Scientific Discovery Engine NovumMind Released: AI Independently Proposes Hypotheses and Discovers New Superconducting Compounds

Autonomous Scientific Discovery Engine NovumMind Released: AI Independently Proposes Hypotheses and Discovers New Superconducting Compounds

On March 19, 2029, the Max Planck Institute for Artificial Intelligence in Berlin announced that its NovumMind system had independently completed the full scientific discovery process — from hypothesis generation to experimental verification — without human intervention. Within three weeks, the system discovered four previously unknown high-temperature superconducting candidate compounds.

NovumMind project lead Professor Thomas Richter stated at the press conference: "This is not an assistive tool — it is a genuine autonomous researcher. The system independently read over 2 million physics papers, autonomously formulated a new hypothesis about copper oxide lattice structures, and then designed 317 experimental protocols."

The system comprises three core modules: a hypothesis generator based on a 12-billion-parameter scientific reasoning model capable of identifying contradictions and gaps in massive literature; an experiment designer that filters viable approaches through physics simulation; and an execution module connected to four automated laboratory devices for material synthesis and electrical measurements.

During its three-week autonomous run, NovumMind executed 284 material synthesis experiments. Four compounds exhibited zero-resistance properties above 95K. An independent verification team at the Technical University of Berlin has confirmed the superconductivity of two of these compounds.

However, the results have sparked intense debate in the academic community. Cambridge University philosophy of science professor Eleanor Marsh noted: "When AI can autonomously discover scientific laws, we need to redefine the attribution of discovery rights and intellectual property. Who gets listed as first author on these papers?"

NovumMind's findings have been submitted to Nature and Physical Review Letters for peer review. The Max Planck Institute has indicated that the system's code and experimental data will be open-sourced after the review process is complete.