Oh boy, the ethics problems do manifest in the real world much more than alignment problems. If only for the simple reason that ethics problems can and do arise

Oh boy, the ethics problems do manifest in the real world much more than alignment problems. If only for the simple reason that ethics problems can and do arise in the use of very simple models (logistic regression) and are not preconditioned on achieving AGI or anything close. For example a logistic regression model for fraud detection that bases it's decision mostly on zip codes. Is actually a proxy for deciding based on race (in the US at least) These kind of models are already massively used and commercially sold not as some curiosities but as decision makers that impact the day to day life's of millions if not billions of ppl.