Aligned to whom?
A paper or article examining the question of whose values AI alignment should serve. Its central argument is that alignment is not a purely technical problem but a question of power distribution, asking which group's values an AI system should ultimately be aligned with.
Timeline (last 90 days)
An article titled Aligned to whom? received 158 upvotes on Hacker News and sparked 95 comments.
The core argument of the article is that alignment is not a purely technical problem but a question of power distribution.
The article recommends that model developers disclose which value judgments they have embedded, treating transparency as a prerequisite for alignment.
The article advocates giving users and society the space to choose among different value configurations, echoing the emphasis on autonomy and controllability in the open-source model community.
All Facts (3)
The core argument of the article is that alignment is not a purely technical problem but a question of power distribution.
50%UnverifiedThe article recommends that model developers disclose which value judgments they have embedded, treating transparency as a prerequisite for alignment.
50%UnverifiedThe article advocates giving users and society the space to choose among different value configurations, echoing the emphasis on autonomy and controllability in the open-source model community.
50%