20160309

OK I guess I just witnessed the history.
AlphaGo definitely wrote the history.
I mean, well, I definitely did not expect to witness this moment.
Though the result is not surprising.
I do not study in the field of AI, but I know some basic ideas. And I know a well-trained AI could be very strong doing its specific task.
Yet Google does reached a height no one else could ever tough.
When IBM first build Deep Blue, the concept of parallel computing they used is still the basic method of doing large-scale computing today. And they also have Waston--another interesting AI which built with a different approach.
But this time, this machine is armed with deep learning and neuron network…And I always believe that neuron network is the right way to lead us to the next stage.
Bottom-up should be a better way--because a top-down knowledge based AI will have many limits. They are essentially restricted by those rules programmed into them, they are not feasible and their intelligence can not exceed the level of who "gave" them their knowledge.
Something that can learn--that is what neuron network does.
Still, we are far from a true AI. I mean, strong AI, of course.
But the point is, we are taking the first step.
Now it is the problem of time, money and maybe a little luck.
I am excited but worried.
As someone stepping into the area, this is a great news, a break through, and they have shown us a way to hope.
But if I say I am not afraid at all, I am lying.
Cause that is something beyond us--something we cannot fully understand anymore.
I remember some guy talking about researchers just fed AI program data without knowing what is actually going on inside. Actually it is a common way to train neuron networks.
But..OK for now it still works. But what if someday we found that the AI is self-evolving and..go beyond us?
Well looks like many people are worrying about a Skynet, but it will not happen..at least for now.
The computers are not fast enough yet, their hardwares are not feasible, and most importantly, they can only do very specific tasks.
The last piece of the puzzle, I believe, will be the quantum computer.
They can solve problems we cannot solve fast enough now, and they have some important properties. For example, the true randomness.
I know Caltech has a team working on this, but it seems we are still far from something practical.
Another article I read before says AI will cause human to either be eternal or be extinguished.
Personally I tend to believe the latter. Just because I tend to be negative about the future.
Although the only thing I believe in is the nature itself..but now humans are changing the law of nature itself.
If we change the rule, what will happen next?
I don't know.
Maybe human will handle this just fine, like we deal with nuke--that could be the best outcome.
But if we made any mistake, then.

Well let us just wait for now.
I kind of believe in fate.
Although quantum mechanics tells us it should not exist.
But it does not give an answer on why things happen.
Maybe someone can find out later--is that determined as well?

Anyway, I am gonna keep on finding my own path.
Good night.

About Eternal Wings

Just another guy on the web.