Previously I asked whether lack of (or slow) belief propagation might actually be a
defense mechanism against harmful beliefs from agents with mal-intent being implanted. It turns out that
even when given common priors, in this experiment
people gave more weight to things they observed or concluded themselves, rather than those communicated by other participants. Via Overcoming Bias; and of direct interest to the decision-theory engineers at Less Wrong.
No comments:
Post a Comment