From the Wikipedia page: "A friendly artificial intelligence... is a hypothetical artificial general intelligence (AGI) that would have a positive rather than negative effect on humanity."

http://en.wikipedia.org/wiki/Friendly_artificial_intelligence

Except it's not really that hypothetical. As sci-fi as it sounds, there are organisations like the Machine Intelligence Research Institute (intelligence.org) working on developing a friendly AI right NOW.

Their general intent seems to be to create a superintelligent program that can and will dedicate itself to solving the problems that an assortment of human brilliant minds haven't figured out yet.

Wanna know how to live forever? This AI will think of a way to eliminate the aging process. Want to play Sword Art Online? It's working on an AmuSphere for you. World hunger, cancer, AIDS, pollution... the theory is that you can simply be smart enough to solve a problem, no matter what. Well, maybe you or I can't, but a program with access to more processing power than a human brain could.

So Dollars, what do you think about it all?

I'd love to have like a cortana

before we get to "superintelligence", we have difficulties enough making "intelligent" programs, don't you think?

a lot of ai that we do currently have hasnt reached the point of understanding language in the same way you and i can. being smart isn't the issue; programming doesn't accomplish "smart" as much as it accomplishes "fast". if we told a really smart person that we want them to make an amusphere, the difficulty comes from sorting through the world's nearly-endless data, discerning the relevant information, and combining the information in a meaningful way given certain constraints. even moreso, what if a solution doesn't exist? at what point would it stop trying?