The Future of ML in Simulation

I think the future of ML in simulation being combined and it opens up new trajectories. You cannot come up with a physics based model, but you could collect a lot of data and derive a model from that. And then you combine basically the best of all worlds. I would potentially always see in software 2.0 and 4.0 or whatever humans adding their creative minds into the models and the software of the future. It's maybe not so easy to imagine how software basically writes itself. But we're getting ideas in, we are setting the boundaries and then the algorithm will do the job.