Do models still matter?

There’s a lot more focus on implementation than back a few years ago when there was POC after POC but far fewer models made it to production. There was also the beginnings of more focus on explainable ML and more focus on scrutinising models for bias of different types. However as MLOps has really taken off the focus on productionising seems to have lead to less focus on the models themselves. Is that still out there or is it just …

bias datascience focus implementation matter mlops poc production types