AI can go bad; no doubt in my mind on that. But because it's code, generally it'll be because a programer didn't account for something or accidentally/intentionally coded bias. The moral dillema about the trolley and diverging tracks, one track you'll run over a loved one the other you run over 5 innocent people, which track do you choose? is a good example. How each person answers is, even with only two choices, is different along with the justifications. Now, an AI is programmed to run our colony trains/cars/vehicles. If it ever encounters a scenario that comes close to a moral dilema, whoever programmed it will have more than likely biased it. Depending where/what you think/justify it might be an agreeable solution, then again, maybe not. So what about a terminator situation? Asgardia has the unique bias afforded to us that, without bloodshed or repressive conflict, we're seeking a future for human kind. More than likely, an AI we give over to that goal, will see the positive side (because it's a positive attempt not born from negative actions) (implicit morality?)so it should behopefully less likely to think extermination is a viable option. Just dont ask me about sentience in machines from AI..... you'll be reading for a month. LOL