I would say probably not, because that’s really hard to do and probably not worth it. Rendering hair physics is indeed very computationally intensive, but that’s not what the AI is really doing, it’s generating frames out of noise and statistical likelihood of pixel values based on training data and previous frames. It’s not really generating hair or cats per se, just what’s likely to be next based on what was before and what was in thousands and thousands of other videos of cats. It kinda bypasses the the need for physics simulation. However, this process too is extremely computationally intensive and expensive for that matter, but typically they’re short low resolution videos generated by online services that are running datacentres to be able to do this and doing so at a massive financial loss while they figure out where the money is going to come from with this emerging tech. It’s both more impressive and easier than hair simulation in some ways and also harder and worse in many ways.