Can legal artificial intelligence achieve reliability and justification?
Peer-Reviewed Publication
Below are the latest research news from Science China Press.
Updates every hour. Last Updated: 3-Nov-2026 02:17 ET (3-Nov-2026 07:17 GMT/UTC)
Artificial Intelligence (AI) can predict judicial decisions, but can it explain how they are legally jusitified, and can judges trust it? A Perspective in National Science Review argues that legal computing requires higher interpretability and reliability, and judicial values should be embedded into model and algorithm design. It outlines a value-guided framework within a “society–law–ethics–algorithms–data” system and identifies three scenario-specific applications: legislation, feedback-validated enforcement and trustworthy, human-centered adjudication. The Perspective further suggests that combining empirical learning with formal reasoning may offer a particularly promising path forward.
A team led by researchers at Peking University Cancer Hospital & Institute reports that combining targeted KRAS inhibition with CTLA-4 immune checkpoint blockade overcomes the immunosuppressive microenvironment that renders KRAS-mutant gastric adenocarcinoma resistant to PD-1 immunotherapy. In a genetically faithful, immunocompetent mouse model of KRASG12D-driven gastric cancer, the combination suppressed TGF-beta signaling, impaired regulatory T cell function and elicited effective anti-tumor immunity. The study, published in Science China Life Sciences, provides a preclinical rationale for CTLA-4-based combination immunotherapy in this molecularly defined subset of patients.
Researchers at Fudan University have developed a crosslinked potassium polyacrylate skin membrane that lets microrobots generate electricity by walking on metal or silicon surfaces. The membrane uses atmospheric moisture and oxygen to convert environmental chemical energy into power, achieving 133 mW cm−2 on zinc. It also senses substrate material and roughness, enabling active energy pursuit. The system powered over 10,000 walking steps and remained stable after 1,000,000 steps.
Rock weathering is an important process regulating atmospheric CO2 concentrations and long-term climate evolution, yet how enhanced erosion affects rock weathering and its carbon effects remains poorly understood. Focusing on the Lancang River basin on the southeastern margin of the Tibetan Plateau, the authors quantitatively reveal the differential responses of silicate, carbonate, and sulfide weathering to changes in erosion. The results show that, as erosion intensifies, silicate weathering rates decrease while sulfide oxidation rates increase, causing weathering in the basin to tend toward a net carbon source. The related findings have now been published in Science China Earth Sciences.
The high O–H bond strength of common alcohols prevents their direct hydrogen atom transfer (HAT) activation. By leveraging the β-silicon effect in α-silyl alcohols, the research group of Professor Xiao Shen at Wuhan University weaken the O–H bond to enable redox-neutral, photoinduced homolysis using thioxanthone. The resulting alkoxy radicals undergo Brook rearrangement to carbon radicals, which are intercepted by nickel catalysis for enantioconvergent cross-couplings with acyl chlorides and aryl bromides. This protocol tolerates substrates inert under classical SET/PCET conditions and establishes direct HAT as a versatile platform.
From rhythmic activity in localized populations of cortical neurons to dengue hotspots that repeatedly reappear across Medellín, Colombia, localized oscillations shape collective dynamics in complex networked systems ranging from the brain to urban epidemics. However, existing explanations are often constrained by strict requirements on system dimensionality. This study reveals a more general mechanism for the emergence of localized oscillations, showing how collective dynamics can transcend these conventional constraints and broadening the theoretical framework for understanding localized collective behavior.
HER2-positive breast cancer accounts for about 15% of all breast cancers, and although anti-HER2 targeted therapies have dramatically improved patient survival, primary and acquired drug resistance remain major clinical challenges. In a comprehensive review, researchers from Fudan University Shanghai Cancer Center, Kunming Medical University, and collaborating institutions map ten major categories of resistance mechanisms, evaluate strategies to overcome them—including next-generation antibody-drug conjugates—and highlight predictive tools that are paving the way toward precision treatment.