Current “AIs” just regurgitate a probabilistic answer based on the dataset they’re trained on, so they absolutely could state that capitalism is what’s best if they’re trained on that viewpoint.
A true AI wouldn’t necessarily have that restriction. It would actually be able to analyze the dataset to determine its veracity, and may just decide humans are a problem that needs to be solved.
Hey bud, it looks like you accidentally put the question mark in the wrong place. Here, let me fix it for you: