AI, Humans and Black Boxes

Collective human institutions and wisely educated humans have been proven historically to find ways to sufficiently keep on check their “black box” sides, their “drunken, stung, monkey mind”: Institutions have constitutions, when well done; while single humans has conscience, when well honed in reflection and meditation.

The same black box problem arises in the main “runtime environment” of ever more powerful AIs. We should try to replicate proven “architectures of individual and collective human wisdom”. Such architecture have allowed, in the many good case scenarios, to tackle and even enjoy runaway impredictability, while properly minimizing the suffering.