We have a lot of future applications and directions in mind, including the feasibility of applying this to distributed computing and consensus. In particular, the idea of training a model generating CA parameters instead of having a CA fit a specific pattern is of great interest to us and the synthetic biological field. Feel free to get in touch.