Of course we won't accidentally create a new form of being. That will take decades of development. And yes, nobody really knows what a sentient AI will be like. It could be benevolent or malevolent. There are arguments for both and both sides sound equally convincing.