RE: AI isn't as smart as you think

There are three AI development stages largely accepted.

ANI
AGI
ASI

ANI - Artificial Narrow Intelligence. This is where we are, if I teach a computer to play Chess, I can't expect it to be able to play Checkers. ANI learns how to do one task very well based on how well you teach it.

AGI - Artificial General Intelligence. This is way in the future, although many claims we are getting close. If I teach an AGI how to be a lawyer, I can expect it to know how to be a doctor. AGI can do most things well, but not necessarily as good as a human, but you don't have to train it on every thing individually.

ASI - Artificial Super Intelligence. This is where many believe may result in the end of the world as we know it. This is where you no longer need to train AI, it is capable of learning on it's own and at a pace that dwarfs a human. Think of how quickly a straight A student can learn advanced calculus compared to how quickly a high school drop out would learn the same subject. This is what is referred to as a snowball effect, once an AI gets 5x smarter than humans, it will quickly get 10x smarter, 100x smarter, and ultimately 1M times.