i just realized that we may need to take a step even further back here
skilled: gto means "Game Theory Optimal"
this is a mathematical subject area initiated by nash and it is designed for deriving optimal strategies in games with thinking/reacting opponents and incomplete information. the purpose of GTO(gto) play is to ensure that no matter what happens, the opponent cannot win by employing an algorithm of any kind whatsoever.
in other words, the result of a gto bot is a bot that successfully implements either: a) a strictly dominant strategy (i.e. one that outperforms all other strategies in expectation, or b) an equilibrium strategy (i.e. one that results in a stalemate in expectation if all expert perfectly playing players play in this fashion)
since a) is basically impossible in poker, and since b) provides all that is needed to generate an expert playing bot, the UofA folks focused on developing a mixed strategy that suuuper closely approximates a nash equilibrium. the result is a bot that plays its strategy so well that it gets close enough to the target nash equilibrium that it is indistinguishable statistically over something like 60million hands. so in effect, it was designed to be unexploitable (i.e. can't lose). THAT'S the purpose of a gto optimal strategy. it's to avoid exploitation no matter how good the opponent is.
hopefully this helps. if you'd like to learn more, you should read how they designed the bot:
http://webdocs.cs.ualberta.ca/~games...ai-cfrplus.pdf