Artifically intelligent physics solver: This AI understands Newtons law
V. Nirmala, Arjun Ram Rajagopal · Science & Technology Journal · 2019
implemented a working prototype of a Deep Learning module that seem to understand Newton’s third law of motion. The networks. In this paper, a Google BERT neural network model was trained using transfer learning technique on a synthetic dataset of simple physics problems within the scope of solving Newton’s third law problems that requires understanding of concepts such as action and reaction, magnitude and direction forces, simple concepts of vectors in physics problems. The of Netwon’s third law assuming certain boundaries on the language model of the word problems. A working prototype of this AI can be accessed at the given website. This paper also contributes the source code for reproducible results. This novel idea can be extended to more science topics. Applications of this interdisciplinary area of AI and physics have impact not just in areas of robotics and computational physics, but also in how science uses AI in the future. In future, more areas of .