It's not enough to characterize every disparate outcome as "systemic racism" and then to "fight systemic racism". We need to talk about what are the specific unjust factors that contribute to these disparities and identify solutions to combat them.

An example of putting forth a concrete, actionable model and proposing a solution might be (and note that this is only an example):

* jurors are more likely to convict a black defendant than a white defendant

* we believe that jurors see racial disparities in crime committance and project them onto defendants

* to combat this, we might try promoting cultural integration to reduce the racial disparities in crime committance as well as advocating for colorblind antiracism so jurors are less likely to see a defendant as a token of his race

The important thing is that the example (1) puts forth a testable model for how the system works and (2) proposes solutions based on that model. You may find yourself disagreeing with our hypothetical model, and that's fine. Note that you can even debate it because it's testable (you can probably conceive of data which rebut it). This is in contrast to esoteric debates about systemic racism, which seem more religious in nature than useful for combatting unjust disparities.