I’m an engineer, so I look at this from a physical sciences point of view. The main problem with the “no free will” argument is it provides no predictive power, there is no model that can say person X will do Y (instead of A, B, C or D) in situation Z.
What is possible is giving probabilities of Y, A, B, C or D in experimental settings. But in the real world, there are too many variables interacting in a chaotic manner to even give reasonable probabilities; this is why we can only use population level statistics rather than individual level predictions.