we once thought only we were capable of.

漸漸的被機械所超越

Today's computers can beat us in complex board games,

現在的電腦可以 在複雜的桌遊中打敗我們

transcribe speech in dozens of languages,

把演講翻譯成各種語言

and instantly identify almost any object.

還能立刻辨識各種物品

But the robots of tomorrow may go futher

在未來，機器人可能會透過

by learning to figure out what we're feeling.

理解我們的感受變得更進步

And why does that matter?

為什麼這很重要？

Because if machines and the people who run them

假如機器人以及操作者 可以精準判斷出我們的情感

can accurately read our emotional states,

可能會給我們很大的幫助， 或是進而操控我們

they may be able to assist us or manipulate us

但在討論到那之前

at unprecedented scales.

試問為什麼像情緒這麼複雜的東西

But before we get there,

可以轉換成 機器唯一理解的語言符號：數字呢？

how can something so complex as emotion be converted into mere numbers,

其實跟大腦先學習認知情緒 才會詮釋自己的心情是一樣的

the only language machines understand?

美國心理學家Paul Ekman 歸類出幾種廣泛使用的情緒

Essentially the same way our own brains interpret emotions,

這些情緒的視覺線索 在各個文化中意義都相同

by learning how to spot them.

舉例來說，一個笑臉圖像 對都市人和原住民來說

American psychologist Paul Ekman identified certain universal emotions

都是代表喜悅

whose visual cues are understood the same way across cultures.

根據Ekman

For example, an image of a smile signals joy to modern urban dwellers

憤怒

and aboriginal tribesmen alike.

厭惡

And according to Ekman,

恐懼

anger,

喜悅

disgust,

悲傷

fear,

以及驚訝都相當有辨識度

joy,

事實證明， 電腦辨識影像的速度愈來愈快

sadness,

多虧了像人工神經網絡 這種學習演算法

and surprise are equally recognizable.

這些人工節點會藉由互相連結 和交換資訊來模仿生物神經的活動

As it turns out, computers are rapidly getting better at image recognition

為了訓練網絡， 會事先輸入預先歸類的樣本，

thanks to machine learning algorithms, such as neural networks.

例如把標示快樂或悲傷的照片輸入系統

These consist of artificial nodes that mimic our biological neurons

網絡會根據特定的特徵調整相關數值， 學習如何辨別這些樣本

by forming connections and exchanging information.

輸入的訓練資料愈多， 運算法辨識新資料就會愈準確

To train the network, sample inputs pre-classified into different categories,

這跟我們的大腦一樣：

such as photos marked happy or sad,

藉由學習以往經驗， 形成往後對新刺激的反應

are fed into the system.

辨識演算法不只會 辨認臉部表情而已，

The network then learns to classify those samples

情緒表達有很多種方式，

by adjusting the relative weights assigned to particular features.

包括肢體語言，語調

The more training data it's given,

心率變化、膚色，以及表皮溫度，

the better the algorithm becomes at correctly identifying new images.

甚至包含說話節奏 和書面的語法結構

This is similar to our own brains,

你可能認為訓練神經網絡 辨識它們是漫長又複雜的程序

which learn from previous experiences to shape how new stimuli are processed.

不過當你了解網路數據的龐大 和現代電腦運算的速度就會改觀了：

Recognition algorithms aren't just limited to facial expressions.

像是從社群網站的發文 到上傳的照片和影片；

Our emotions manifest in many ways.

從通話紀錄到熱感應監視器；

There's body language and vocal tone,

還有記錄生理狀況的穿戴式裝置

changes in heart rate, complexion, and skin temperature,

最大的問題已經不是如何取得資料，

or even word frequency and sentence structure in our writing.

而是使用這些資料的目的

You might think that training neural networks to recognize these

電腦化表情辨識有很多實用途徑：

would be a long and complicated task

利用演算法辨識表情的機器人

until you realize just how much data is out there,

可以幫助孩童學習

and how quickly modern computers can process it.

或是陪伴孤單的人

>From social media posts,

社群媒體公司正在考慮運用演算法

uploaded photos and videos,

來標記含有特定文字或用語的發文 以協助自殺防治

and phone recordings,

情緒辨識軟體 可以幫助治療心理疾病

to heat-sensitive security cameras

甚至提供人們低價位的 自動化心理治療

and wearables that monitor physiological signs,

儘管有這些好處，

the big question is not how to collect enough data,

大型網路自動掃描我們的照片、 對話紀錄、和生理層面是相當惱人的

but what we're going to do with it.

當我們的情緒數據被公司利用來打廣告 我們該如何保有隱私？

There are plenty of beneficial uses for computerized emotion recognition.

同時也出現人權問題

Robots using algorithms to identify facial expressions

警方能夠把未決定犯罪的人 直接判定為罪犯嗎？

can help children learn

機器人的發展還有很長的路要走，

or provide lonely people with a sense of companionship.

例如辨別像諷刺這種細微的情緒

Social media companies are considering using algorithms

以及分辨情緒強弱， 例如一個人多快樂或多難過

to help prevent suicides by flagging posts that contain specific words or phrases.

儘管如此，機器人將來 可能會精準地判斷情緒

And emotion recognition software can help treat mental disorders

並做出回應

or even provide people with low-cost automated psychotherapy.

但是人類會不會因 機器人的過度入侵感到恐懼，

Despite the potential benefits,

這又是另一回事了

the prospect of a massive network automatically scanning our photos,

communications,

and physiological signs is also quite disturbing.

What are the implications for our privacy when such impersonal systems

are used by corporations to exploit our emotions through advertising?

And what becomes of our rights

if authorities think they can identify the people likely to commit crimes

before they even make a conscious decision to act?

Robots currently have a long way to go

in distinguishing emotional nuances, like irony,

and scales of emotions, just how happy or sad someone is.

Nonetheless, they may eventually be able to accurately read our emotions

and respond to them.

Whether they can empathize with our fear of unwanted intrusion, however,

that's another story.