TheAncientGeek comments on The case for value learning

[TheAncientGeek]

To be fair to Bostrom’ he doesn’t go all the way down the rabbit hole—arguing that oracles aren’t any different to agentive AGIs.