Does AI really have the potential to take over?

In Mara Hvistendahl's article "Can we stop AI outsmarting humanity", she talks about the dangers of AI and the precautions necessary to deal with it. She highly references Jaan Tallinn in the article who is an Estonian born computer programmer that wonders what AI could do if it gets too powerful. He doesn't really fear the thought of a single AI or an army of AI, but he does fear about what it does and how it does it. He continuously pours money into  foundations working on AI safety and does so to make sure he is supporting one that sticks around. If he worries about it this much and he's not alone, it makes us wonder; can or will AI ever take over the human race?

For AI to take over the human race, we would have to be very careless on how we use it and how we develop it to become a powerful AI. First of all, an AI has to be questioning things to even get it's knowledge. Or at least programmed to question. If it isn't programmed to question, it can really only retain the information given to it, unless there is an uncontrollable variable like a glitch in the system. Another way AI could become powerful is if it wasn't fully developed, but if it wasn't developed fully, could it really seek knowledge? I don't think it can because it won't have the necessary "tools" to get what it needs.

All in all, i don't think that AI will take over. I simply think that we will get addicted to the point where it seems to "control us". An AI would have to do a lot to get out of its physical and programmable barrier. And from what it seems, programmers and computer engineers have made so many that it would take a lot to do that.