ADM-201 dump PMP dumps pdf SSCP exam materials CBAP exam sample questions

4个AI实验和惊喜! – 译学馆
未登录,请登录后再发表信息
最新评论 (0)
播放视频

4个AI实验和惊喜!

4 Experiments Where the AI Outsmarted Its Creators

Dear Fellow Scholars, this is Two Minute Papers with Károly Zsolnai-Fehér.
亲爱的学霸们 这里是Károly Zsolnai-Fehér为您带来的《两分钟论文》
Today, I am really excited to show you four experiments
今天 我非常兴奋地要向你们展示四个实验
where AI researchers were baffled by
在实验中 AI研究人员被他们自己发明的机器人
the creativity and unexpected actions of their own creations.
所展现的创造力和意想不到的行为所困扰
You better hold on to your papers.
你最好抓紧你的论文哦
In the first experiment,
在第一个实验中
robots were asked to walk around
机器人被要求四处走动
while minimizing the amount of foot contact with the ground.
同时要最小化足部与地面的接触面积
Much to the scientists’ surprise,
令科学家们惊讶的是
the robots answered that this can be done with 0% contact,
机器人回应可以零接触完成这个任务
meaning that they never ever touch the ground with the feet.
意思是它们根本不需要用脚接触地面
The scientists wondered how that is even possible,
科学家们想知道机器人是怎样做到的
and pulled up a video of the proof.
并拿出了一个证据视频
This proof showed a robot flipping over and walking using its elbows.
证据显示机器人翻转了过来 用它的肘部行走
Talk about thinking outside the box!
就是我们说的“在盒子外思考”!
Wow.

A different robot arm experiment also came to a surprising conclusion.
在另一个机器人手臂实验中也得到了令人惊讶的结论
At first, the robot arm had to use its grippers to grab a cube,
首先 机器人手臂必须用它的手爪抓取一个立方体
which it successfully learned to perform.
它成功地学会了抓取立方体
However, in a later experiment, the gripper was crippled,
然而 在之后一次实验中 手爪出了故障
making the robot unable to open its fingers.
使得机器人无法打开它的爪子
Scientists expected a pathetic video with the robot to push the box around
科学家们期待看到机器人四处推着箱子
and always failing to pick up the cube.
但总是无法捡起立方体的可怜视频
Instead, they have found this.
然而 他们发现了这个
You see it right, instead of using the fingers,
你看到了吧 不是使用的爪子
the robot finds the perfect angle to smash the hand against the box
机器人找到了一个完美的角度 用手爪猛击盒子
to force the gripper to open and pick up the box.
强制把手爪打开 然后捡起盒子
That is some serious dedication to solving the task at hand.
这对于解决突发状况是非常重要的贡献
Bravo!
好样的!
In the next experiment,
在下一个实验中
a group of robots were tasked to find food
一组机器人被要求去寻找食物
and avoid poisonous objects in an environment,
同时要避开环境中的有毒物品
and were equipped with a light and no further instructions.
虽然配备了灯光 但却没有更多指示
First, they learned to use the lights to communicate
它们首次学会了使用灯光来互相交流
the presence of food and poison to each other and cooperate.
食物和有毒物品存在的位置 并且学会了合作
This demonstrates that
这个实验证明了
when trying to maximize the probability of the survival of an entire colony,
当尝试最大化整个群体的存活可能性时
the concept of communication and cooperation can emerge even from simple neural networks.
甚至在简单神经网络中也会产生交流和合作的概念
Absolutely beautiful.
简直美极了
And what is even more incredible, is that later,
更令人难以置信的是随后发生的事情
when a new reward system was created that fosters self-preservation,
当一种促进自我保护的 新的奖励机制被创造出来时
the robots learned to deceive each other by lighting up the the food signal near the poison
它们学会了通过点亮有毒物品附近的食物信号来互相欺骗
to take out their competitors and increase their chances.
以此来打败竞争对手和增加自己的生存机会
And these behaviors emerge from a reward system and a few simple neural networks.
这些行为产自于一个奖励系统和一些简单的神经网络
Mind blowing.
让人脑洞大开
A different AI was asked to fix a faulty sorting computer program.
另外一个人工智能被要求去修复一个计算机程序分类的错误
Soon, it achieved a perfect score without changing anything
很快 在没有改变任何东西的情况下 它就获得了一个完美的分数
because it noticed that by short circuiting the program itself,
因为人工智能注意到通过程序短路本身
it always provides an empty output.
它总是输出空值
And of course, you know,
当然了 你是知道的
if there are no numbers, there is nothing to sort.
如果没有数字 也就没有东西需要分类
Problem solved.
问题解决了
Make sure to have a look at the paper,
请一定要看一下这篇论文
there are many more experiments that went similarly,
还有更多类似的实验
including a case where the AI found a bug
其中有一个例子是AI在一个物理仿真程序里
in a physics simulation program to get an edge.
找到了一个漏洞以取得优势
With AI research improving at such a rapid pace,
随着人工智能研究的飞速发展
it is clearly capable of things that surpasses our wildest imagination,
它显然是有能力超越我们最疯狂想象的事物
but we have to make sure to formulate our problems with proper caution,
但我们必须确保以谨慎的态度 来陈述我们的问题
because the AI will try to use loopholes
因为人工智能会尝试使用漏洞
instead of common sense to solve them.
而不是常识去解决这些问题
When in a car chase,
在一场汽车追逐中
don’t ask the car AI to unload all unnecessary weights to go faster,
不要为了让汽车人工智能跑得更快 而叫其卸下所有不必要的重量
or if you do, prepare to be promptly ejected from the car.
如果你这么做 那就做好立马被甩离汽车的准备
If you have enjoyed this episode,
如果你喜欢本集视频
please make sure to have a look at our Patreon page in the video description
请一定要在视频说明中里查看我们的Patreon页面
where you can pick up really cool perks,
在那里你可以获得很酷的特权
like early access to these videos
比如提前观看这些视频
or getting your name shown in the video description, and more.
或在视频说明中显示你的名字等等
Thanks for watching and for your generous support,
感谢您的观看和慷慨支持
and I’ll see you next time!
我们下次再见!

发表评论

译制信息
视频概述

AI在其飞速的发展过程中总是能超过实验者的想象,让我们一起来看一下这四个实验中AI的惊人表现吧!

听录译者

收集自网络

翻译译者

三十四

审核员

审核员XY

视频来源

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

相关推荐