Fair point. But of course the reasons for the nonlinearities are different. ML is not inherently nonlinear; ML is only nonlinear because linear neural nets are not very useful.

On the other hand linear control theory is quite useful for many real-world problems.