True AI Going Rogue

[Crab Puncher]

Do you feel AI is going to be a problem in the near future? I know I have some quarries with it. what about you?

[AskTJ]

Doubt it.

[Beef Boss]

Don't really care, I mean so what if it does we terminate it. 

[AlexTheGreatish]

I think it could be a problem.  Not in the way where an AI develops malice towards humans, but instead we accidentally create something that is orders of magnitudes smarter than us and it accidentally wipes us all out.  I feel like there could be an issue right at the start, when a general AI is born that is both smarter than any person, but also doesn't know much about humans and is dedicating all of it's power towards one problem.

 

Like if we were to create a general AI and task it with trying to find extraterrestrial life.  Instead of worrying about humans, it might build a bunch of robots that harvest materials to build space ships, and accidentally wipes out most humans in the process, much like how we will not even think about a large colony of ants or squirrels when building a city.

[Eaglerino]

Could just unplug it

[Mira Yurizaki]

The problem is that we as humans don't have anything else to compare to when it comes to higher level thinking. The closest we may have something to compare to are animal-communication experiments like Koko the Gorilla.

 

In my view about the rogue AI scenario is that AI has to grow up like we do. Think of it this way: if you were to raise a human in effective isolation from the outside world until say 25, with the math and science training you'd expect someone a reasonably educated 25 year old to have (let's assume at least basic college level) and then suddenly thrust them into the outside world without guidance, how do you think they would behave? And you're expecting this person to behave like a normal person. Essentially AI should be treated as something that is very smart, but with zero experience with how the world works. So if you want AI to "behave", you're going to have to allow it to experience the world and make decisions on its own and on its own terms.

 

Also in another way of putting it, we would be thrusting our moral compass onto something else that can determine its own moral compass. A rogue AI is only rogue because it came to a very logical conclusion that doesn't align with our moral compass. If it decides to solve a human generated problem by killing all humans, that's a very logical approach. It just doesn't agree with our moral compass (generally speaking)

 

In any case, until nanomachine colonies are a thing, robots are pretty fragile things if you think about it. A car may be able to kill me by going fast, but a car has more than a dozen easily exploitable weakpoints to render it useless. A Roomba can't really do anything. Even if an AI controlled airplane goes nuts, what's it going to do? Crash into something?

[JCBiggs]

It will most likely go rogue.. but unless it figures out how to exist in the ether, and not inside our electronics, it will never NOT be under our control.  and if it does figure out how to exist in the ether, then that means something else probably already has to an that will keep it in line.  

 

Im with timstips.  I think genetic programming is going to be the biggest fiasco we have to deal with right now.   Can you imagine even every peice of pollen on the planet had some kind of genetic engineering that caused other plants to get sick and die?    its not as far fetched as you would think.