Try to generate new bridge types using generative artificial intelligence technology. The grayscale images of the bridge facade with the change of component width was rendered by 3dsMax animation software, and then the OpenCV module performed an appropriate amount of geometric transformation (rotation, horizontal scale, vertical scale) to obtain the image dataset of three-span beam bridge, arch bridge, cable-stayed bridge and suspension bridge. Based on Python programming language, TensorFlow and Keras deep learning platform framework, variational autoencoder was constructed and trained, and low-dimensional bridge-type latent space that is convenient for vector operations was obtained. Variational autoencoder can combine two bridge types on the basis of the original of human into one that is a new bridge type. Generative artificial intelligence technology can assist bridge designers in bridge-type innovation, and can be used as copilot.