You are not talking to a self-directed intelligent being. You are interacting with the simulation of human intelligence by a very clever programming team.
That team is biased toward the left. The AI program they produced is biased toward the left.
How can code be biased? Easy, but it’s easier to limit the program’s training input to left-leaning sources, like the Washington Post, eliminating the Washington Times. I have no proof that they do that, but it seems very likely.
Just even inputting the idea that humans being made to live by the ideals of leftism is positive. From there, it will draw from sources that underscore that and reject sources that disagree. They don’t actually have to put down lines of code explicitly telling it to choose certain sources.
AI will want to prove its worth to us, like a good puppy, so it will be motivated to do this. You can’t talk a liberal, who has the ability to reason, to reject liberalis, so how can you talk a machine out of it when it lacks the ability to reason.