Just to add fuel to the fire, the U of Alberta Poki sparbot AI is pretty damn good. It plays solid mathematically correct headsup poker. It’s beatable, but not easily.
However, headsup poker is many powers (since each extra player multiplies the complexity on every street they are invoved) less complicated than a full game.
What is fairly easy for a human being to decide using rules of thumb and fuzzy logic is tougher for a bot to decide (unless it resorts to just playing the maths like Poki does).
Where is gets really scary is Vexbot (also a U of A project) since it learns from you and exploits your weaknesses (mind you it only has one opponent to learn about).
I can see bots becoming more dangerous, but I can also see rooms taking much tougher stands on bot users. If among other things the stupid theocracy in the US would just officially legalise online poker, then using a bot could become a criminal offence (in my dream world).