MiguelDev comments on Which possible AI systems are relatively safe?

[MiguelDev]

or the conceptual architecture of corrigibility is a part of the attention mechanism...