Artificial Intelligence (AI) and machine learning offer a wide variety of new computational methods and techniques to further improve and complement model-based image reconstruction methods. Such tools include the use of neural field representations of images and novel training algorithms and hybrid reconstruction methods incorporating learned reconstruction methods. This works develops a range of AI-assisted model guided image reconstruction methods and applies them for image reconstruction in Photoacoustic computed tomography (PACT) and ultrasound computed tomography (USCT). This results in computationally efficient image reconstruction methods for PACT and USCT and allow them to transfer into widespread adoption.