Nick Bostrom: What happens when our computers get smarter than we are?
尼克.波斯特羅姆: 當電腦比人類聰明時會發生什麼?
Nick Bostrom asks big questions: What should we do, as individuals and as a species, to optimize our long-term prospects? Will humanity’s technological advancements ultimately destroy us? Full bio
Double-click the English transcript below to play the video.
philosophers and computer scientists,
及電腦科學家一起工作。
the future of machine intelligence,
思考機器智慧的未來。
things are sort of science fiction-y,
這是科幻小說的範疇,
human condition.
現代人類的狀況....
guests on this planet,
was created one year ago,
would be 10 minutes old.
two seconds ago.
看看這一萬年來的GDP增長
world GDP over the last 10,000 years,
to plot this for you in a graph.
for a normal condition.
這是個很有趣的形狀。
of this current anomaly?
「是什麼造成了這種異態呢?」
through human history,
科技在人類歷史上不斷累積,
advances extremely rapidly --
so very productive.
我們現在的生產力很高。
to the ultimate cause.
回想到最終的原因
distinguished gentlemen:
tokens, an incredible feat.
這是一個難以置信的壯舉。
superstring revolution.
他掀起了第二次超弦革命。
this is what we find:
這是我們看到的:
in the exact way it's wired.
be too complicated, however,
been 250,000 generations
take a long time to evolve.
需要很長的時間演化。
to intercontinental ballistic missiles.
到發射洲際彈道飛彈
至今我們所實現的所有事
that everything we've achieved,
changes that made the human mind.
is that any further changes
the substrate of thinking
enormous consequences.
think we're on the verge
a profound change in that substrate,
about putting commands in a box.
將指令輸入到一個箱子裡。
handcraft knowledge items.
for some purposes,
you couldn't scale them.
你無法延展這些系統。
what you put in.
僅限於你事先輸入的範圍。
in the field of artificial intelligence.
around machine learning.
representations and features,
often from raw perceptual data.
的能力的程式碼。
that the human infant does.
limited to one domain --
某個領域的人工智慧 --
between any pairs of languages,
在任何兩種語言之間翻譯
on the Atari console.
的任何一款遊戲。
the same powerful, cross-domain
還未能達到像人類一樣
as a human being has.
algorithmic tricks
how to match in machines.
複製到機器中。
to match those tricks?
在機器裡面複製這些技巧?
leading A.I. experts,
做了一次問卷調查,
and one of the questions we asked was,
其中的一個題目是:
there is a 50 percent probability
人類會有50%的機率
human-level machine intelligence?"
as the ability to perform
定義為有能力
as an adult human,
像一名成年人一樣好,
within some limited domain.
而不是僅限於某些領域。
group of experts we asked.
much later, or sooner,
也有可能提早實現
limit to information processing
機器基體處理資訊能力的最終界限
the limits in biological tissue.
at 200 hertz, 200 times a second.
可能在200赫茲,每秒200次。
operates at the Gigahertz.
都以千兆赫(GHz)的頻率運轉。
100 meters per second, tops.
比較慢,頂多是每秒100公尺。
at the speed of light.
to fit inside a cranium,
必需要放得進顱骨內。
of a warehouse or larger.
跟倉庫一樣大,甚至更大。
lies dormant in matter,
現在正潛伏在物質裡面,
lay dormant throughout human history,
在人類的歷史中一直潛伏著,
the power of artificial intelligence.
將人工智慧的力量喚醒。
an intelligence explosion.
見證到智慧的大爆發。
about what is smart and what is dumb,
什麼是聰明什麼是愚笨的時候,
roughly like this.
會是這樣的:
or whoever your favorite guru is.
或你喜歡的某位大師。
of artificial intelligence,
probably more like this:
at zero intelligence,
years of really hard work,
辛苦的研究以後,
mouse-level artificial intelligence,
老鼠級的人工智慧,
cluttered environments
of really hard work, lots of investment,
及投資了很多資源之後,
chimpanzee-level artificial intelligence.
黑猩猩級的人工智慧。
of really, really hard work,
artificial intelligence.
we are beyond Ed Witten.
我們就超越了愛德維騰。
at Humanville Station.
在人類村這一站就停車。
to questions of power.
twice as strong as a fit human male.
比一個健康的男性人類要強壯兩倍。
and his pals depends a lot more
則很大的部分取決於
what the chimpanzees do themselves.
而非黑猩猩們自己的作為。
on what the superintelligence does.
超級智慧的作為。
that humanity will ever need to make.
最後一個發明。
at inventing than we are,
on digital timescales.
做出這些事。
a telescoping of the future.
that you could have imagined
in the fullness of time:
可以發明出來:
of minds into computers,
或將我們的頭腦上載到電腦裡,
with the laws of physics.
develop, and possibly quite rapidly.
而且速度可能很快。
technological maturity
it would be able to get what it wants.
得到它想要的東西。
be shaped by the preferences of this A.I.
被這個超級智慧的偏好所影響。
what are those preferences?
這些偏好是什麼呢?
avoid anthropomorphizing.
將機器智慧擬人化(人格化)。
every newspaper article
每一篇關於未來的人工智慧
has a picture of this:
to conceive of the issue more abstractly,
更抽象的來想像這個議題,
as an optimization process,
一個優化的過程,
into a particular set of configurations.
特定的組態的過程。
a really strong optimization process.
是一個很強大的優化過程。
available means to achieve a state
conenction between
是有意義的目標之間
would find worthwhile or meaningful.
to make humans smile.
是讓人類笑。
or amusing actions
它會做出有用的或是好笑的動作
effective way to achieve this goal:
可以達到這個目標:
muscles of humans
a difficult mathematical problem.
解出一個非常困難的數學問題。
to get the solution to this problem
into a giant computer,
一部超大號的電腦,
an instrumental reason
會給人工智慧理由去做
might not approve of.
problem from being solved.
的過程中成為阻礙。
go wrong in these particular ways;
事情不會以這種方式出錯;
optimization process
非常強大的優化流程
that your definition of x
in many a myth.
教導的寓意。
he touches be turned into gold.
他碰到的所有東西都可以變成金子。
she turns into gold.
她變成了黃金。
他的食物也變成了黃金。
一個強大的優化流程
optimization process
不正確或不精確的目標後
or poorly specified goals.
sticking electrodes into people's faces,
開始在人臉上安裝電極,
if we've grown dependent on the system --
已經對這個系統產生依賴性 ——
to the Internet?
flicked the off switch to humanity,
沒有把人類的開關關掉?
for example, right here.
an intelligent adversary;
and plan around them.
並為其做出準備。
at that than we are.
that we have this under control here.
覺得一切都在我們的掌握之中。
a little bit easier by, say,
把AI放到一個盒子裡面
from which it cannot escape.
the A.I. couldn't find a bug.
這個AI不會找到漏洞?
find bugs all the time,
他們還經常找出漏洞。
to create an air gap,
製造一個物理間隙,
using social engineering.
來突破物理間隙。
out there somewhere
有一名公司職員
her account details
的人士說服(詐騙)
from the I.T. department.
around in your internal circuitry
can use to communicate.
用以與外界溝通。
you up to see what went wrong with you,
把你打開檢查哪裡出錯時,
to a really nifty technology,
很巧妙的科技藍圖,
that the A.I. had planned.
秘密副作用。
not be confident in our ability
對我們控制人工智慧的能力
locked up in its bottle forever.
只是時間問題而已。
is to figure out
such that even if -- when -- it escapes,
哪怕是它逃出來了,
fundamentally on our side
因為它是站在我們這一邊的
this difficult problem.
that this problem can be solved.
a long list of everything we care about,
我們在乎的所有事物寫下來,
in some computer language
寫成電腦程式語言
that uses its intelligence
一個人工智慧,它用它自己的智慧
in such a way that it is motivated
會讓它想要
that it predicts we would approve of.
去做它認為我們會贊成的事情。
its intelligence as much as possible
利用到它們的智慧
very good for humanity.
可對人類是非常有益的。
for the intelligence explosion
這個智慧的大爆炸,
in just the right way
need to match ours,
要和我們的一致,
how the A.I. behaves,
很簡單低檢查它的行為,
that the A.I. might encounter
可能會遇到的情況下
that would need to be solved, sorted out:
uncertainty and so forth.
邏輯不確定性的情況等問題。
to be solved to make this work
a superintelligent A.I.,
那樣的難,
is a really hard challenge.
是一個很難的挑戰。
challenge on top of that.
how to crack the first challenge
有人想出了如何解決第一個難題
the additional challenge
work out a solution
by the time it is needed.
我們可以用的到它。
the entire control problem in advance
完全解決「控制」的問題
can only be put in place
你所想要控制的架構後
architecture where it will be implemented.
that we solve in advance,
事先解決更多的難題
to the machine intelligence era
機器智能時代的機率
that is well worth doing
things turn out okay,
look back at this century
回顧我們這個時代的時候
the one thing we did that really mattered
所做的最重要的事就是
ABOUT THE SPEAKER
Nick Bostrom - PhilosopherNick Bostrom asks big questions: What should we do, as individuals and as a species, to optimize our long-term prospects? Will humanity’s technological advancements ultimately destroy us?
Why you should listen
Philosopher Nick Bostrom envisioned a future full of human enhancement, nanotechnology and machine intelligence long before they became mainstream concerns. From his famous simulation argument -- which identified some striking implications of rejecting the Matrix-like idea that humans are living in a computer simulation -- to his work on existential risk, Bostrom approaches both the inevitable and the speculative using the tools of philosophy, probability theory, and scientific analysis.
Since 2005, Bostrom has led the Future of Humanity Institute, a research group of mathematicians, philosophers and scientists at Oxford University tasked with investigating the big picture for the human condition and its future. He has been referred to as one of the most important thinkers of our age.
Nick was honored as one of Foreign Policy's 2015 Global Thinkers .
His recent book Superintelligence advances the ominous idea that “the first ultraintelligent machine is the last invention that man need ever make.”
Nick Bostrom | Speaker | TED.com