Actually computers that learn by their own mistakes have worked for some years now. Especially I remember a computer, which was running a “baby program”. The computer could form its own messages and communicate with the scientists, who built it. The program was designed so that it would evolve slowly, pretty much like a human baby. It started out by putting words together randomly, but slowly it learned how to make childish sentences.
I psychologist with speciality in children were given a print of the computer’s conversation with one of the scientists. The psychologist said that it looked like a perfectly normal child at about five years old.
But it is quite new to me that robots can build themselves and learn from their mistakes. This is actually quite cool!
And no, I don’t believe the robots ever will be able to take the power from us humans. I’m sure they could do it if they were allowed to, but scientists will always make it possible to turn them off if something should happen.

guy, you don't seem to be able to wrap your head around what's happening in the field of AI, or biology, for that matter.

First of all, humans use every part of their brains. Not at the same time, however. This is the part that confuses people. At any given time, you are only using a small percentage of your brain. This makes intuitive sense to me. We probably understand more about the brain than you think. It is only a matter of time before we learn the last of the brain's secrets.

As for AI, there's no reason we can't recreate the human brain in the form of a machine. I don't think we'd have to do that, personally, to create subversive artificial intelligence, but I'm certain it could be done.

Thank you clagnol for making clear that we do use all of our brains, just not at the same time. That is the part that did sort of get me (I thought we might use it all at different times, but to use it all at once has not yet been achieved).

I do understand the world of the AI and biological, and I tend to try and keep up-to-date with such things.

Secondly, several reports I have read, been told and seen on the television say that we have only unlocked a small percentage of our brain. I may not have made my point clearly though, this is what I meant: We may know what drives our brains and how they work (as in the sense of electro-chemical pulses) however, we do not know how powerful our brain can be. Many scientists argue that the human brain never stops increasing in 'power'. Unlike computers, it has no clocked speed. And every brain is different. And as for the 'never' being able to, let me rephrase: Not in the near future. And following on from clagnol's last point, why would we? If the scare is that high about robots conquering humans, then they will never be designed to carry out such actions.

Like clagnol said, it's 20% of our brain at any given moment in time.
It'd be like an appendix. If we don't use it, why not just cut it out? No, we don't do that because every part is useful!

Just taking a intro course to Psychology, you'll see that there are many different parts in charge of different things. They don't say "Well, this is the 10-20% we use, and the rest is grey matter that just sits there."

This is an amazing breakthrough for science! I never thought I'd see the day that we built a robot capable of learning and correcting its mistakes. I'm sure that eventually we will build more robots, like in the movie "I-Robot" (except hopefully not the part where thy go evil and try to destroy the human race as we know it )

.Solitude

This isn't real AI because these things can't learn or make their own thoughts. The things are programmed to check if there are mistakes then reform if there are.

No, the programms are given a set of open ended variables. From that, the collection of data within the system can "work things out". AI is not about having thoughts, to have thoughts is to live. Computers can never essentially live, as long as they are artificial.

It's the beginning of Judgement Day. This is pretty cool though. I wonder how far we can take this to? I wouldn't mind sharing the world if these robots contribute too

I mean real AI has the ability to learn and do things that weren't originally programmed. In other words, individual thought. In a way I guess. Use code they create for themselves rather than what was created for them.

If the robots come together to create a graphing/scientific calculator that fixes *my* mistakes, I'll be the happiest college student ever.

Seriously though ... interesting. I don't see how robots would get the idea to take over the world - seeing as they don't have thought processes, so to speak - but wow, what if we had a real RoboCop going around and taking out known terrorists? That would be cool.