Several years ago, I worked on the technical privacy and security design and architecture for systems that used ML for prescriptions and some diagnostic triage.

Several years ago, I worked on the technical privacy and security design and architecture for systems that used ML for prescriptions and some diagnostic triage. At the time, I thought this was too big an ethical issue to be pushed down to us technologists to solve (as are many in government), so I got in touch with an eminent philosopher about the question of whether a person was recieving honest and ethical care from ML and from whom? (15 years ago now). His response to my (limited and naive) question was essentially, people will hold attachments to beliefs about the human element in these transactions, and the philosophical part of the question was why people would hold beliefs that were not strictly rational, and he sort of declined the implied ethical questions. There was no reason to expect him to respond with more, but given I was navigating the ethics of AI driven medical care via questions of privacy in system design decisions (read: accountability) for the institutions who would use it and the millions of people subject to it, it seemed like an opportunity to be at the very forefront of what would likely become a defining social issue of our lives in a couple of short decades. What we discovered then as architects, as most people are just about to, is that the main use case for ML/AI will be to use complexity to diffuse accountability away from individuals, and aggregate it up into committees and ultimately corporate entities that are themselves artificial beings. AI is the corporeal manifestation of an institution, essentially a golem.