奥创 天网 格拉多斯
Ultron, Skynet, GLaDOS.
These are wonderful characters, we really do love them.
And they should be fine, but films and games
involving an artificial super intelligence as a rule commit a cardinal sin.
They are unreasonable and ultimately unrealistic.
What they portray cannot be.
And it’s not just that is fiction and artificial intelligence should be defeated.
传染病 战争 天灾 核武器 全球变暖……
Epidemics, wars, natural disasters, nuclear weapons, global warming…
Mankind on our beautiful blue planet has always had to face a variety of problems.
And each time we have climbed out of the hole we’ve dugged for ourselves,
a new threat has appeared on the horizon and it is a terrible thing that
so many scientists call every latest danger
the greatest threat to humanity in all of history.
And occasionally, our very last invention.
Do you know what are the differences between us and ants?
The effect is ants cannot think any human fashion.
They cannot perceive of the existence of other worlds,
the immensity of the universe, or the depth of the ocean.
They may know in their own trivial way
that there are such things as skyscrapers or people.
but they will never comprehend that skyscrapers are built by people.
And do you know how we treat ants?
We stomp on them.
Once machines learn to think and improve themselves
and when these machines can reason millions of times faster than human being,
they might master in 5 minutes what would take a human 10 years.
If no one presses the stop button, they might become something like gods
in less time than it takes to blink an eye.
We will not just be ants to them.
They will surpass us a million times over.
They will be able to calculate billions of possibilities and probabilities,
揭示所有自然规律 通晓意识的奥秘 甚至摧毁世界
reveal every law of nature, understand the secrets of consciousness, even destroy the world.
If they want, that is.
The smarter a machine becomes, the faster it would increase its own intelligence.
It may take years for a machine to surpass ants’ intelligence.
but possibly in a minute for them to become smarter than us.
And that other artificial intelligence would become smarter and smarter with every passing second.
There will suddenly be something on this earth
much higher than us on the ladder of evolution,
极其高端的机器 如果它们想要 就能像碾压蚂蚁一样碾压我们
infinitely higher and it would crush us like ants if it wants to.
It will all come down to luck.
Most often when we envision an artificial super intelligence we make a grave mistake.
We imagine it would be some kind of good and friendly computer,
a helpful assistant who would always support and assist us.
But this is not so.
It would be something completely alien to us.
And at the same time able to reason.
如果我给你一只毛茸茸的豚鼠 并告诉你 她不咬人
If I hand you a big fluffy guinea pig and tell you she doesn’t bite.
You would pat her.
But let’s say I offer you instead a giant spider that is 400 times smarter than you.
Still want to pat him? Of course not.
A giant smart tarantula might be able to carry on an interesting conversation.
But he would still be a spider,
a cold-blooded killer with the instincts of an insect which we do not understand at all.
In films, artificial super intelligence is generally portrayed as something humanlike.
所以 也许对我们来说 当那一天真的来到 一切都不会有问题
And so it might seem to us that when that day really comes everything would be all right.
We are already accustomed to it as something
like a standard old villain from horror movies, slightly clumsy and toothless.
But a real artificial intelligence when faced with the task like making people smile,
might for example, paralyze people’s facial nerves to keep them smiling permanently.
One wrong task like this and the planet would be full of ever-smiling people,
but with eyes filled with horror.
One could ask an artificial intelligence to end hunger on the planet.
它会说 很简单 没问题 然后它就会消灭所有人类 这样也就消灭了饥荒
and it would say “that’s easy, no problem” and then just kill everyone or hunger.
Set this AI with the task to preserve life on Earth as much as possible.
And they could easily kill all humans again.
Imagine that a machine that has super intelligence
which produces cars in a factory suddenly goes haywire.
It has just this one task–to produce as many cars as possible.
It improves on its own algorithms.
It shuts the factory to the outside world so that no one can interfere with its mission.
It redesigns and rebuilds the assembling machines,
so that they work more quickly and efficiently.
It organises the supply of necessary materials which are immediately assembled
by robots that obey only the super intelligence.
In no time at all the world might be covered in the grey dusty devices of our all life.
And eventually the planet itself would be broken down into raw material.
Just to fabricate more and more cars.
And who knows what might happen next?
This insane AI could even go to other planets
in search of raw material to continue to do its job.
This could all happen so quickly that no one would even have a chance to pull the plug.
What to do? Nobody knows yet.
This is not like dodging a meteorite. This is much more serious.
The creation of artificial intelligence really could be the last invention of mankind.
奥创 天网 格拉多斯