"Foresee"? "Coming"? The example you mentioned is already with us. Insurance companies and lenders already use computerized models with problematic outcomes, which are completely opaque: https://scholarworks.law.ubalt.edu/cgi/viewcontent.cgi?artic...
"AI" is just speeding up processes that started in the 1970s.
"Nobody knows why the machine learning module denied black people a mortgage 43% more frequently than whites, but it's 'AI' shrug"
One of the biggest priorities as we shape this future needs to be a way not only for the algos to make correct decisions, but also the ability for us to interrogate the decision-making process so we can be proactive about the kind of future we want this technology to give us.
Because it is coming, without a doubt.