Universal basic income isn’t always AGI-proof — LessWrong