Tag: #ai-for-science22 results
Alibaba DAMO Academy, in collaboration with academic institutions, launched the AI agent 'ElementsClaw'. The AI predicted 68,000 potential superconductors, with 4 new materials already experimentally verified.
Researchers utilized AI to identify four previously unknown superconductors in just 28 GPU hours. This breakthrough significantly accelerates the discovery process compared to traditional human-led methods.
The U.S. Department of Energy selected 278 projects for its Genesis Mission, a national AI research initiative backed by more than $250 million initially and potentially over $5 billion in future funding. Projects span underground microbiology, critical-mineral recycling, and autonomous nuclear-reactor design, but researchers question the short timelines and low selection rate.
AI for Science is rapidly moving toward closed-loop systems that generate hypotheses, run experiments, verify results, and feed validated findings back into the next cycle. The article warns that these systems may improve throughput while weakening human scientists’ taste and judgment, and may remain trapped within patterns present in existing data.
Jeff Dean and other prominent Google executives have founded Discovery Loop, a new AI startup. The company will pursue AI-driven breakthroughs in areas including drug discovery and chip design.
USTC and the Beijing Humanoid Robot Innovation Center launched Labimus, a simulation and evaluation platform for humanoid robots in precision chemistry labs. It uses a 3x4 evaluation matrix to test robot precision and robustness in tasks like powder weighing and tool handling.
YaoSu Tech has secured nearly 200 million RMB in A+ funding to advance its '3D Bio Intelligence' platform. The funds will support AI model development, high-throughput automation, and the creation of virtual cell and biological world models.
London-based AI lab Inherent has emerged from stealth with $50 million in seed funding. The startup aims to build AI systems capable of identifying high-value scientific research questions.
Sugon has deployed China's largest 60,000-card AI for Science (AI4S) computing cluster at Zhengzhou's national supercomputing node. It features ultra-high compute, full-precision support, high-speed networking, and applications accelerating simulations in biology and materials by orders of magnitude. Integrated with OneScience platform for one-stop AI4S model development.