In the field of Large Language Models(LLMs) vertical applications, the innovative direction of deeply integrating foundational LLMs with industrial scenarios has emerged. Researchers have developed an intelligent decision support system for CNC system fault diagnosis, utilizing LLMs and domain knowledge graph(KG) technologies. This advancement effectively overcomes the limitations of typical expert systems in symbolic reasoning efficiency and accuracy. The study, published in Engineering, provides insights and methods for the practical application of industrial large models.

New research from the University of Georgia is shedding light on how different countries are preparing for how AI will impact their workforces.

Artificial intelligence (AI) can effectively support doctors in making diagnoses. It makes different mistakes than humans—and this complementarity represents a previously untapped strength. An international team led by the Max Planck Institute for Human Development has now systematically demonstrated for the first time that combining human expertise with AI models leads to the most accurate open-ended diagnoses. 

A novel power supply technology for 3D-integrated chips has been developed by employing a three-dimensionally stacked computing architecture consisting of processing units placed directly above stacks of dynamic random access memory. Towards realizing this, researchers developed key technologies involving precise and high-speed bonding technique and adhesive technology. These new technologies can help in addressing the demands of high-performance computing applications, which require both high memory bandwidth and low power consumption with reduced power supply noise.  

Photothermoelectric (PTE) photodetectors with self-powered and uncooled advantages have attracted much interest due to the wide application prospects in the military and civilian fields. However, traditional PTE photodetectors lack of mechanical flexibility and cannot operate independently without the test instrument. Herein, we present a flexible PTE photodetector capable of dual-mode output, combining electrical and optical signal generation for enhanced functionality. Using solution processing, high-quality MXene thin films are assembled on asymmetric electrodes as the photosensitive layer. The geometrically asymmetric electrode design significantly enhances the responsivity, achieving 0.33 mA W^-1 under infrared illumination, twice that of the symmetrical configuration. This improvement stems from optimized photothermal conversion and an expanded temperature gradient. The PTE device maintains stable performance after 300 bending cycles, demonstrating excellent flexibility. A new energy conversion pathway has been established by coupling the photothermal conversion of MXene with thermochromic composite materials, leading to a real-time visualization of invisible infrared radiation. Leveraging this functionality, we demonstrate the first human–machine collaborative infrared imaging system, wherein the dual-mode photodetector arrays synchronously generate human-readable pattern and machine-readable pattern. Our study not only provides a new solution for functional integration of flexible photodetectors, but also sets a new benchmark for human–machine collaborative optoelectronics.