AI safety as systems engineering versus an alignment problem
One post contrasts an engineering view built around clear trade-offs with an alignment argument: you can't make trade-offs on something you don't understand or control.
TLDR
A post argues that a “huge disconnect” separates people who treat AI safety as systems engineering with clear trade-offs from those who approach it as an alignment problem, questioning trade-offs without understanding or control. A reply tentatively agrees, asks which side the author takes, and presses for how the engineering view explains what it calls “recent alignment incidents.”
Combined views
398
2 Sources, first seen ago
