未登录,请登录后再发表信息
最新评论 (0)
播放视频

人工智能的最终目标

Inside Google's DeepMind Project: How AI Is Learning on Its Own | Max Tegmark

我简单地将智能定义为是对
I define intelligence simply as how good something
实现复杂目标很有帮助的东西
is at accomplishing complex goals.
现在的人工智能与机器智能在许多方面
Human intelligence today is very different from machine intelligence today
都有着极大的区别 首先 以前的机器智能
in multiple ways. First of all, machine intelligence in the past
仅仅用于低级的人类智能
used to be just in always inferior to human intelligence.
渐渐地 机器智能在一些极其有限的领域
Gradually machine intelligence got better than human intelligence in
甚至比人类更优秀 像便携计算机算乘法一样快
certain very, very narrow areas, like multiplying numbers fast like pocket calculators
或是迅速记忆大量数据
or remembering large amounts of data are really fast.
如我们所见 相比过去狭窄的尖端领域
What we’re seeing now is that machine intelligence is spreading out
机器智能只是运用的范围更广了一点
a little bit from those narrow peaks and getting a bit broader.
现在仍然没有东西能和人类的智慧一样包罗万象
We still have nothing that is as broad as human intelligence,
正如我们所知 一个人类小孩
which you know where a human child can learn
能习得各种技能完成几乎所有的目标
to get pretty good at almost any goal.
但已经有程序能学习如何
But you have systems now for example that can learn to
玩一系列的不同电脑游戏
play a whole swath of different kinds of computer games
或是学会在极其复杂的环境中驾车
or to learn to drive a car and in pretty varied environments.
并且 人工智能的范围在明显拓宽
And where things are obviously going in AI is increased breadth
而人工智能研究的圣杯是要造出一个机器
and the holy grail of AI research is to build a machine.
它像人类智能一样擅长任何事
that is as broad as human intelligence can get good at anything.
一旦实现 人工智能极有可能
And once that’s happened it’s very likely
不仅达到像人类领域一样宽泛 甚至在
it’s not only going to be as broad as humans, but also better than humans
所有的任务完成上超过人类 与现在只是一些截然不同
at all tasks, as opposed to just some right now.
我必须承认我是一个电脑迷
I have to confess that I’m quite the computer nerd myself.
我在高中和大学时写了一些电脑游戏
I wrote some computer games back in high school and college
而最近我在麻省理工的实验室进行
and more recently I’ve been doing a lot of
大量的深度学习研究
deep learning research with my lab at MIT.
所以一些事真的让我发出“哇”的赞叹
So something that really blew me away like “whoa”
当我第一次看到谷歌的能从零开始学
was when I first saw this Google deepmind system
玩电脑游戏深入学习系统时
that learned to play computer games from scratch.
有这样一个人工模拟神经网络
You had this artificial simulated neural network,
它不知道电脑游戏 电脑 屏幕都是什么
it didn’t know what a computer game was it didn’t know what a computer was, it didn’t know what a screen was
你只需输入代表屏幕上不同颜色的数字
you just fed in numbers that represented the different colors on the screen
告诉它 它能输出不同的数字
and told it that it could output different numbers
对应着它仍然无法理解的不同的按键
which correspond to the different keystrokes it also didn’t know anything about.
然后一直传送数据 而据我所知这些尝试
And then kept just feeding it the score and all this offer I knew
是为了随机地做出可以得到最大值的东西
was to try to randomly do stuff that would maximize that score.
你会说我记得在屏幕上看到谷歌深入学习的的CEO戴密斯·哈萨比斯展示过这些
You say well I remember watching this on the screen once when Demis Hassabis the CEO of google deepmind show them,
看到前半场这个东西完全遵守英国策略并且一直输
seeing the first half this thing really played total BS strategy and lost all the time,
慢慢地就变得越来越好然后变得比我以前还好
gradually got better and better, and then it got better than I was
然后过一会它就发现了打砖块的疯狂策略
and then after a while it figured out this crazy strategy in Breakout
(你要让球在一片砖墙下弹跳)
(you’re supposed to bounce the ball off of a brick wall)
它一直对准左上角直到在那戳出一个洞
where it would keep aiming for the upper-left corner until it punched a hole through there
然后让球在上面跳来跳去得到大量分数
and got the ball bouncing around in the back and just racked up crazy many points.
然后我就感叹 哇 它真聪明
And I was like, “Whoa, that’s intelligent!”
甚至编码这个程序的人也不知道这个策略
And the guys who programmed this didn’t even know about that strategy
因为他们不怎么玩那个游戏
because they hadn’t played that game very much.
这是机器智能可以超过它的创造者智能的
This is a simple example of how machine intelligence can surpass the intelligence of its creator,
一个简单例子 就像人类小孩
much in the same way as a human child can
在接受良好教育后变得比父母
end up becoming more intelligent than its parents
更有智慧是一样的 并且
if educated well. And
这只是小的普通电脑
this is just tiny little computers
这种硬件在你笔记本上也可能有
the sort of hardware you can have on your desktop.
现在想象一下按比例增加到世界上最大的
If you now imagine scaling up to the biggest computer facilities we have in the world
电脑并给我们这几十年的算法发展
and you give us a couple of more decades of algorithm development,
我认为我们极有可能造出一些机器
I think it is very plausible that we can make machines
它不仅能学会如何比我们玩游戏玩的更好
that cannot just learn to play computer games better than us
还会将生活视为游戏
but that it can view life as a game
做什么都比我们好
and do everything better than us.

发表评论

译制信息
视频概述

通过主人公在谷歌的见闻谈到人工智能的发展,人工智能在机器学习领域的深入,极有可能使机器在许多方面超越人类

听录译者

收集自网络

翻译译者

朦胧星海

审核员

w

视频来源

https://www.youtube.com/watch?v=X2tr0lEmslw

相关推荐