My feeling is that until a carbon life form can be distinguished from a vote bot there will be no trust in Proof of Brain.
One solution was explored in an old blog post of mine, whose link is included below for your reading pleasure.
https://steemit.com/steem/@novacadian/a-decentralized-capcha-annonymous-proof-of-brain-verification
How is it that there are massive number of bots on reddit, twitter, youtube, etc. and still the curation system works far better (not perfect) than it does on Steem? Bots are subject to the same economic incentives as humans (remember, every bot is ultimately run by a human). If humans can make money via stupid or abusive voting then so can bots an vice versa. Indeed I would say that Steem is (or at least should be with the right economic incentives) less susceptible to manipulation by bots because raw vote count has a lesser to negligible role, and mass voting is one way in which bots have clear advantage over humans (try voting with 10k sock puppets without a bot).
Yesterday was my first opportunity to edit a past payout posting. No sooner had it been updated than esteem recognized 10 100% 25 rep account votes from someone's bygone experiment. They must be triggered on post url.
There are plenty of dumb bots that's for sure.