Can robots be creative? - Gil Weinberg

翻译人员: Zongzhen Yang 校对人员: Zhiqiang Pi

How does this music make you feel?

这音乐让你感觉如何？

Do you find it beautiful?

你觉得好听吗？

Is it creative?

有创意吗？

Now, would you change your answers

那么如果现在告诉你作曲家 是一个机器人，

if you learned the composer was this robot?

你会改变刚才自己的回答吗？

Believe it or not,

信不信由你，

people have been grappling with the question of artificial creativity,

人们一直在试图攻克人工创意

alongside the question of artifcial intelligence,

及人工智能这两大难题，

for over 170 years.

为此研究了一百七十多年。

In 1843, Lady Ada Lovelace,

1843年，Ada Lovelace夫人

an English mathematician considered the world's first computer programmer,

一位被誉为世界上第一个 电脑程序员的英格兰数学家

wrote that a machine could not have human-like intelligence

写道机器不会有人类一样的智慧。

as long as it only did what humans intentionally programmed it to do.

如果人们只让机器按照所编程程序它们的，

According to Lovelace,

根据Lovelace，

a machine must be able to create original ideas

一个机器必须要能够创造新的想法

if it is to be considered intelligent.

才算是聪明。

The Lovelace Test, formalized in 2001, proposes a way of scrutinizing this idea.

2001年形成的Lovelace测试提出了一个检测这个想法的方法。

A machine can pass this test if it can produce an outcome

如果机器能够形成程序员无法根据原代码所解释的

that its designers cannot explain based on their original code.

输出才可以通过检测。

The Lovelace Test is, by design, more of a thought experiment

根据设计，Lovelace测试更像是思想实验

than an objective scientific test.

而非客观的科学检测。

But it's a place to start.

但是这是一个起点。

At first glance,

乍一看。

the idea of a machine creating high quality, original music in this way

机器创作高质量，原版的音乐的想法

might seem impossible.

看起来不可能。

We could come up with an extremely complex algorithm

我们可以想出一个使用随机生成数字，

using random number generators, chaotic functions, and fuzzy logic

混乱的函数和模糊的逻辑来创造一系列音符的，

to generate a sequence of musical notes

极为复杂的算法，

in a way that would be impossible to track.

使其不可能理解。

But although this would yield countless original melodies never heard before,

但是即使这样能够产生无数的，没人听过的原版旋律，

only a tiny fraction of them would be worth listening to.

只有一小部分值得一听。

With the computer having no way to distinguish

电脑是不法区分

between those which we would consider beautiful

我们觉得好听

and those which we won't.

或者不好听的旋律。

But what if we took a step back

但是如果我们退一步

and tried to model a natural process that allows creativity to form?

然后尝试塑造一个自然的创意形成过程会怎样？

We happen to know of at least one such process

我们恰巧知道至少一个

that has lead to original, valuable, and even beautiful outcomes:

可以带来原创，宝贵甚至美丽的结果：

the process of evolution.

进化。

And evolutionary algorithms,

算法的进化，

or genetic algorithms that mimic biological evolution,

或者模仿物种进化的基因算法，

are one promising approach

是有希望让

to making machines generate original and valuable artistic outcomes.

机器产生原创的，珍贵的艺术品。

So how can evolution make a machine musically creative?

因此计划是如何让一个机器音乐创新？

Well, instead of organisms,

相比生物，

we can start with an initial population of musical phrases,

我们可以从最初的一些乐段

and a basic algorithm

和一个基础的

that mimics reproduction and random mutations

通过调换有些部分，结合另一些

by switching some parts,

和替换一些随机的音节

combining others,

来模仿生育和随机变异的

and replacing random notes.

算法来完成。

Now that we have a new generation of phrases,

现在我们有一个新的音段创作，

we can apply selection using an operation called a fitness function.

我们可以使用一个叫适应函数的运算。

Just as biological fitness is determined by external environmental pressures,

就像生物的适应是由外界压力所导致，

our fitness function can be determined by an external melody

我们的适应函数可以由音乐人或歌迷们

chosen by human musicians, or music fans,

所选择的外界旋律所定，

to represent the ultimate beautiful melody.

以此来表现最终的，动听的旋律。

The algorithm can then compare between our musical phrases

算法可以比较我们的乐段和

and that beautiful melody,

动听的旋律

and select only the phrases that are most similar to it.

从而选择最相似的乐段。

Once the least similar sequences are weeded out,

一旦最不相似的组合被淘汰，

the algorithm can reapply mutation and recombination to what's left,

算法可以再次替换和组合剩下的，

select the most similar, or fitted ones, again from the new generation,

在从新的组合选择最接近的，或者最适合的，

and repeat for many generations.

然后在许多的组合重复。

The process that got us there has so much randomness and complexity built in

这个过程有那么多的随机性和复杂性

that the result might pass the Lovelace Test.

以至于结果可能可以通过Lovelace测试。

More importantly, thanks to the presence of human aesthetic in the process,

更重要的，多亏在这个过程中人类的审美，

we'll theoretically generate melodies we would consider beautiful.

我们理论上可以产生我们觉得动听的旋律。

But does this satisfy our intuition for what is truly creative?

但是这能满足我们对真正创新的直觉吗？

Is it enough to make something original and beautiful,

这足够产生原创而美好的东西吗？

or does creativity require intention and awareness of what is being created?

创新需要意图和意识吗？

Perhaps the creativity in this case is really coming from the programmers,

也许这样的创新其实源于程序员，

even if they don't understand the process.

即使他们不能理解过程。

What is human creativity, anyways?

什么才算是人类的创新？

Is it something more than a system of interconnected neurons

是什么比一个互相连接的，

developed by biological algorithmic processes

由生物算法过程演变的神经元系统

and the random experiences that shape our lives?

以及随机发生的，塑造我们生活的经验还要多的吗？

Order and chaos, machine and human.

秩序与混乱，机器与人类。

These are the dynamos at the heart of machine creativity initiatives

现在有些创意机器的核心

that are currently making music, sculptures, paintings, poetry and more.

正在创造音乐，雕塑，图画，诗歌等等。

The jury may still be out

评委可能会质问

as to whether it's fair to call these acts of creation creative.

是否因当把这些创造行为叫做有创意。

But if a piece of art can make you weep,

但是如果一枚艺术品能让你落泪，

or blow your mind,

能让你大开眼界，

or send shivers down your spine,

或让你全身一颤，

does it really matter who or what created it?

谁创造的还重要吗？