Yes, the problem in all these discussions is that unless you engage with them on broader epistemological grounds, you’re just occluding the a priori biases. En

Yes, the problem in all these discussions is that unless you engage with them on broader epistemological grounds, you’re just occluding the a priori biases. Engineering, as a discipline, has an unfortunate history of not wanting to engage with the social and political conditions under which its work occurs, but that becomes entirely untenable for how a lot of ML is put to use (if you’re trying to be honest anyway).