Cmon, you everyday see AI mistakes. Replacing stuff with AI will be shoot themselves. AI can work with statistics, that`s all. And you have to spend lots of money to teach it before it would be possible to use it. And anyway you have to check its results.
PS. I don`t know is it truth or not, but i read some article not long ago. It told that there are special marks for AI, like "just believe it, don`t need proves" and AI use such marks and ignores doubtful results.
Artificial intelligences have not yet reached a state in which we could call them "perfect", in fact due to them being capable of commiting mistakes is why I believe casinos and banks won't integrate AIs in areas which are directly related with the management of assets or wiring money, handling large withdrawals or deposits, system security an such things, because human beings are still more reliable in many things when compared to AI.
Though, there are areas in which AI will indeed have more advantages like in marketing departments, creation of content costumer service and analyzing odd behaviors in gamblers to flag then or not as possible violators of terms of service.
It is still a long way/path before human beings are fully replaced by thinking machines, in my opinion.
Yes, about "only statistics" i was mistaken. But anyway you have to check the AIs` result. You have to watch his pictures, read his text, etc. It can make your job more comfortable, more easier, but it would be only human assistant. And i don`t think that it would change soon.