The Weekly Brief
Enjoying this newsletter? Know someone who would too? Forward this email to them!
Sign up (it’s free)
December 19th 2025
🦾 Robotics
Chinese authorities unveiled a robot-friendly urban demonstration zone in Shenzhen, allowing humanoid robots to train directly in real-world city environments rather than closed labs. The initiative, part of Guangdong’s “1+1+N” embodied intelligence framework, supports large-scale, scenario-based training for service, manufacturing, and public-sector robots. By combining policy backing, industrial scale, and live urban testing, the program aims to accelerate China’s humanoid robotics and AI deployment.
🤖 Artificial Intelligence
Tech startup Atomic Canyon has developed the first artificial intelligence system specifically trained on nuclear industry data, using the Frontier supercomputer at Oak Ridge National Laboratory - the world’s most powerful exascale machine. The new AI, called the Neutron platform, uses retrieval-focused models trained on over 3 million documents from the U.S. Nuclear Regulatory Commission to help nuclear facilities quickly interpret complex regulations and historical data. The platform dramatically reduces time spent on licensing and safety compliance, boosting productivity and aiding the nuclear sector’s expansion.
Engineers from Stanford University, Carnegie Mellon University, the University of Pennsylvania, and MIT, working with SkyWater Technology, built a multilayer 3D chip that vertically integrates memory and computing to overcome the long-standing “memory wall.” Tests and simulations show up to ten-fold performance gains over conventional 2D chips and a realistic path toward 100–1,000× improvements in speed-energy efficiency for AI workloads, including models related to Meta’s LLaMA. The work, presented at the IEEE International Electron Devices Meeting (IEDM), marks the first commercially fabricated 3D chip to demonstrate clear, large-scale performance advantages.
🌎 Sustainability
🇯🇵 Japanese researchers develop plant-based plastic that dissolves in seawater without microplastics
Scientists led by Takuzo Aida at RIKEN created a cellulose-based plastic that remains strong during use but rapidly decomposes in seawater without forming microplastics. Published in Journal of the American Chemical Society, the material uses salt-sensitive molecular bonds that break apart naturally in marine environments while matching the strength and flexibility of conventional plastics. The innovation could significantly reduce ocean plastic pollution by preventing long-lasting microscopic debris.
💊 Healthcare
Physicians at Children’s Hospital of Philadelphia, working with researchers from the University of Pennsylvania, treated a baby with a fatal metabolic disorder using a one-patient CRISPR gene-editing therapy. The breakthrough allowed the child to leave the hospital, reach normal developmental milestones, and begin walking, marking a new frontier for personalized medicine. His case is shaping future approaches to rare-disease gene therapies and regulatory approval.
Neuroscientists at MIT’s Picower Institute found that briefly anesthetizing the retina of an amblyopic eye can restore visual responses in adult mice by reactivating weakened brain circuits. Published in Cell Reports, the study reveals that temporary retinal inactivation triggers burst-like signaling in visual pathways that helps rebuild lost connections. The findings suggest a potential new treatment strategy for adult lazy eye, a condition long thought to be irreversible after childhood.
Researchers at Mayo Clinic developed an aptamer-based method to selectively label senescent “zombie” cells, a long-standing challenge in aging and disease research. Published in Aging Cell, the team showed that short, shape-shifting DNA molecules can bind unique surface proteins on senescent mouse cells, enabling precise detection without harming healthy cells. The approach, sparked by a graduate student’s idea, could eventually support targeted therapies for aging, cancer, and neurodegenerative diseases.
Researchers at the Icahn School of Medicine at Mount Sinai created an AI system called V2P that links genetic mutations directly to the diseases they are most likely to trigger, going beyond simply labeling mutations as harmful. Published in Nature Communications, the model accurately ranks disease-causing variants using real patient data, helping speed up genetic diagnosis and guide personalized treatment and drug discovery. The advance brings precision medicine closer by connecting DNA changes to specific disease outcomes.
That’s all for today, please reply to this email if you have any comments or feedback, we’d love to hear from you about what we can do better!
Have you enjoyed this email? Make sure to share it with your friends and colleagues.
See you soon,
Max