There is no specific page for something that I’ve seen else-where. It seems to be deliberate AI-enabled lock-in of the existing AI leadership positions, as a substitute for de-jure political authority:
It is being presented as a meta-mistake of alignment, made by assuming that human preferences are static and discoverable, and not having a mechanism for groups of relatively ordinary humans to seek revision of the preferences that the AI is implementing.
There is no specific page for something that I’ve seen else-where. It seems to be deliberate AI-enabled lock-in of the existing AI leadership positions, as a substitute for de-jure political authority:
https://www.notesfromthecircus.com/p/garbage-in-garbage-out
It is being presented as a meta-mistake of alignment, made by assuming that human preferences are static and discoverable, and not having a mechanism for groups of relatively ordinary humans to seek revision of the preferences that the AI is implementing.