On November 30th, Nature published two heavyweight studies: the latest artificial intelligence (AI) driven platform GNoME (Material Exploration Graph Network) has been able to discover and synthesize new inorganic compounds on its own, including the discovery of over 2.2 million stable structures and the creation of 41 new materials in just 17 days. Its speed and accuracy far exceed those of humans. Technological progress has improved the ability of computer programs to recognize new materials, but the main obstacle faced in this process is how learning algorithms can adapt to results that are opposite to what they have learned, as new discoveries are essentially the ability to understand data in new and creative ways. The "Deep Thinking" team has proposed a computational model that can improve the efficiency of material discovery through large-scale active learning. This program uses existing literature to train and generate a variety of potential compound candidate structures, and then continuously improves these structures through a series of learning. GNoME has discovered over 2.2 million stable structures, increasing the accuracy of structural stability prediction to over 80%. When predicting components, the accuracy of every 100 experiments has been improved to 33%, compared to only 1% in previous work. In the second study, a team from the University of California, Berkeley developed an Automated Laboratory (A-Lab) system. This A-Lab is trained based on existing scientific literature, and then combined with active learning, can create up to 5 initial synthesis formulas for proposed compounds. Subsequently, it can perform experiments using a robotic arm to synthesize compounds in powder form. If the yield of a formula is less than 50%, A-Lab will adjust the formula to continue the experiment, and end after successfully achieving the goal or exhausting all possible formulas. After 17 days of continuous experiments, A-Lab conducted 355 experiments and produced 41 out of 58 proposed compounds (71%). In contrast, human researchers need to spend several months guessing and experimenting. The training of AI demonstrated by two research institutes, combined with the rapid development of computing power and existing literature, demonstrates that using learning algorithms to assist in the discovery and synthesis of inorganic compounds has extremely broad prospects. In the future, autonomous laboratories will be able to discover new materials with the least amount of manpower and the fastest speed. (Lai Xin She)