I think macro economics has provided significant value to society even if the model isn't precise with perfect predictions. We do have worthwhile models that the government can use to avoid making serious monetary mistakes in the face of the real business cycle. We are unlikely to trigger the awful depths of the great depression again.

The main problem I see with public economists is Goodharts law: When the predictive model becomes visible to the agents in the economy, there are stable strategies to counteract those models. So if economists want to remain good predictors of real outcomes, they have to collect massive amounts of data to feed the model but then remain mostly invisible afterward, so that the agents being analyzed don't change behavior too much.

In some sense, a known bad model that is used to set central policy may be good because the rest of the economic agents can work on a diversity of better models without correlating as much with each other and the known central model.