正文翻譯
原創(chuàng)翻譯：龍騰網(wǎng) http://www.ltaaa.com 翻譯：roroho 轉(zhuǎn)載請(qǐng)注明出處


See how an AI system classifies you based on your selfie
ImageNet Roulette will take a stab at categorizing you, and it will fail

看看人工智能系統(tǒng)是如何根據(jù)你的自拍照對(duì)你進(jìn)行分類的
ImageNet Roulette將嘗試對(duì)你進(jìn)行分類，而且失敗了

Modern artificial intelligence is often lauded for its growing sophistication, but mostly in doomer terms. If you’re on the apocalyptic end of the spectrum, the AI revolution will automate millions of jobs, eliminate the barrier between reality and artifice, and, eventually, force humanity to the brink of extinction. Along the way, maybe we get robot butlers, maybe we’re stuffed into embryonic pods and harvested for energy. Who knows.

現(xiàn)代人工智能因其日益成熟而經(jīng)常受到稱贊，但大多說(shuō)的是世界末日論。如果你瀕臨于世界末日，人工智能革命將使數(shù)以百萬(wàn)計(jì)的工作實(shí)現(xiàn)自動(dòng)化，消除現(xiàn)實(shí)和技巧之間的障礙，并將最終迫使人類走向滅絕的邊緣。一路走來(lái)，也許我們會(huì)有機(jī)器人管家，也許我們會(huì)被塞進(jìn)胚胎莢里，吸收能量。誰(shuí)知道呢。

“When we first started conceptualizing this exhibition over two years ago, we wanted to tell a story about the history of images used to ‘recognize’ humans in computer vision and AI systems. We weren’t interested in either the hyped, marketing version of AI nor the tales of dystopian robot futures,” Crawford told the Fondazione Prada museum in Milan, where Training Humans is featured. “We wanted to engage with the materiality of AI, and to take those everyday images seriously as a part of a rapidly evolving machinic visual culture. That required us to open up the black boxes and look at how these ‘engines of seeing’ currently operate.”

“兩年多前，當(dāng)我們第一次構(gòu)思這個(gè)展覽時(shí)，我們就想講一個(gè)關(guān)于用計(jì)算機(jī)視覺(jué)和人工智能系統(tǒng)來(lái)“識(shí)別”人類圖像的發(fā)展歷史的故事?！笨藙诟５赂嬖V米蘭普拉達(dá)基金會(huì)博物館（“訓(xùn)練人類”項(xiàng)目就在這里形成）說(shuō)：“我們對(duì)人工智能的大肆炒作、市場(chǎng)營(yíng)銷以及反烏托邦機(jī)器人未來(lái)的故事都不感興趣。我們想?yún)⑴c人工智能的實(shí)質(zhì)性工作，并把那些日常圖像作為快速發(fā)展的機(jī)器視覺(jué)文化的一部分來(lái)認(rèn)真對(duì)待。這就要求我們能打開(kāi)黑匣子，看看這些“視覺(jué)引擎”目前是如何運(yùn)轉(zhuǎn)的?！?br/>
It’s a worthy pursuit and a fascinating project, even if ImageNet Roulette represents the goofier side of it. That’s mostly because ImageNet, a renown training data set AI researchers have relied on for the last decade, is generally bad at recognizing people. It’s mostly an obxt recognition set, but it has a category for “People” that contains thousands of subcategories, each valiantly trying to help software do the seemingly impossible task of classifying a human being.

這是一個(gè)值得追求、有吸引力的項(xiàng)目，即使ImageNet Roulette 代表了其更愚蠢的一面。這主要是因?yàn)镮mageNet，這個(gè)人工智能研究人員過(guò)去十年一直依賴的著名的訓(xùn)練數(shù)據(jù)集，通常不善于識(shí)別人像。它主要是一個(gè)對(duì)象識(shí)別集，但是它有一個(gè)“人”的類別，其中包含成千上萬(wàn)個(gè)子類，每個(gè)子類都積極地嘗試以幫助軟件完成似乎不可能完成的任務(wù)，對(duì)一個(gè)人進(jìn)行識(shí)別分類。

And guess what? ImageNet Roulette is super bad at it.

猜猜怎么著？ImageNet Roulette非常不擅長(zhǎng)這個(gè)。

I don’t even smoke! But for some reason, ImageNet Roulette thinks I do. It also appears to believe that I am located in an airplane, although to its credit, open office layouts are only slightly less suffocating than narrow metal tubes suspended tens of thousands of feet in the air.

我根本就不抽煙！但出于某種原因，ImageNet Roulette卻認(rèn)為我抽。而且它好像還也以為我是在一架飛機(jī)上，盡管值得點(diǎn)贊的是，開(kāi)放式辦公室的布局比掛在幾萬(wàn)英尺高空中令人窒息的狹窄金屬管好那么一點(diǎn)點(diǎn)。

ImageNet Roulette was put together by developer Leif Ryge working under Paglen, as a way to let the public engage with the art exhibition’s abstract concepts about the inscrutable nature of machine learning systems.

ImageNet Roulette是由帕格倫旗下的開(kāi)發(fā)者萊夫·雷奇設(shè)計(jì)的，是一種讓公眾參與藝術(shù)展覽的抽象概念的方式，使他們能了解機(jī)器學(xué)習(xí)系統(tǒng)不可思議的本質(zhì)。

Here’s the behind-the-scenes magic that makes it tick:

以下就是魔術(shù)幕后的秘密，正是它們令其發(fā)揮作用：

ImageNet Roulette uses an open source Caffe deep learning frxwork (produced at UC Berkeley) trained on the images and labels in the “person” categories (which are currently ‘down for maintenance’). Proper nouns and categories with less than 100 pictures were removed.

ImageNet Roulette 使用的是開(kāi)源的Caffe深度學(xué)習(xí)框架（由加州大學(xué)伯克利分校開(kāi)發(fā)），該框架用于“人”的類別（目前在“停機(jī)維護(hù)”）的圖像和標(biāo)識(shí)訓(xùn)練。通過(guò)不到100幅圖片對(duì)專有名詞和類別進(jìn)行剔除。

ImageNet contains a number of problematic, offensive and bizarre categories - all drawn from WordNet. Some use misogynistic or racist terminology. Hence, the results ImageNet Roulette returns will also draw upon those categories. That is by design: we want to shed light on what happens when technical systems are trained on problematic training data. AI classifications of people are rarely made visible to the people being classified. ImageNet Roulette provides a glimpse into that process – and to show the ways things can go wrong.

ImageNet包含的許多有問(wèn)題的、攻擊性的和奇怪的類別——都是從英語(yǔ)詞典WordNet上獲取的。其中有些使用的是厭惡女性或種族主義的術(shù)語(yǔ)。因此，ImageNet Roulette的結(jié)果也將借鑒這些類別。這是故意設(shè)計(jì)成這樣的：我們想弄清楚當(dāng)技術(shù)系統(tǒng)使用有問(wèn)題的訓(xùn)練數(shù)據(jù)進(jìn)行訓(xùn)練時(shí)會(huì)發(fā)生什么。人工智能對(duì)人的分類很少讓被分類的人看到。ImageNet Roulette使我們得以對(duì)這一過(guò)程略窺一二——而且表明這樣做事情就可能會(huì)出錯(cuò)。

ImageNet is one of the most significant training sets in the history of AI. A major achievement. The labels come from WordNet, the images were scraped from search engines. The 'Person' category was rarely used or talked about. But it's strange, fascinating, and often offensive.
— Kate Crawford (@katecrawford) September 16, 2019

ImageNet 是人工智能歷史上最重要的訓(xùn)練集之一，是一項(xiàng)重大的成就。這些標(biāo)簽來(lái)自WordNet英語(yǔ)詞典，這些圖像是通過(guò)搜索引擎搜羅過(guò)來(lái)的。“人”這一類別很少被使用或談?wù)?。但這很奇怪、很吸引人，而且常常令人不快。
——?jiǎng)P特·克勞福德（@katecrawford）2019年9月16日

Although ImageNet Roulette is a fun distraction, the underlying message of Training Humans is a dark, but vital, one.

盡管ImageNet Roulette 是一種有趣的消遣方式，但其“訓(xùn)練人類”項(xiàng)目所傳遞出的潛在信息卻是一個(gè)黑暗但至關(guān)重要的信息。

“Training Humans explores two fundamental issues in particular: how humans are represented, interpreted and codified through training datasets, and how technological systems harvest, label and use this material,” reads the exhibition descxtion “As the classifications of humans by AI systems becomes more invasive and complex, their biases and politics become apparent. Within computer vision and AI systems, forms of measurement easily — but surreptitiously — turn into moral judgments.”

“‘訓(xùn)練人類’項(xiàng)目旨在探索兩個(gè)基本問(wèn)題：如何通過(guò)訓(xùn)練數(shù)據(jù)集來(lái)表現(xiàn)、解釋和編碼人類，以及（人工智能）技術(shù)系統(tǒng)是如何收獲、標(biāo)記和使用這種材料的，”展覽描述中寫(xiě)道，“隨著人工智能系統(tǒng)對(duì)人類的分類變得更具攻擊性和復(fù)雜性，它們的偏見(jiàn)和政治就變得明顯。在計(jì)算機(jī)視覺(jué)和人工智能系統(tǒng)中，進(jìn)行測(cè)量的形式很容易變成道德判斷——但很隱蔽?！?br/>

評(píng)論翻譯
原創(chuàng)翻譯：龍騰網(wǎng) http://www.ltaaa.com 翻譯：roroho 轉(zhuǎn)載請(qǐng)注明出處


YCSMD
"PIPE SMOKER" LMAO
Its not talking about your nicotine usage…

“吸煙斗煙的煙鬼”，哈哈哈笑死寶寶了。
它不是在說(shuō)你吸了多少尼古丁……

Jazzwall
I just tried. It classified my face as: "person, individual, someone, somebody, mortal, soul > face" LOL. It thinks I am a human and have a face!

我只是盡力了。它把我的臉?lè)诸悶椋骸叭?，個(gè)人，某人，某某人，凡人，仙人>臉”哈哈哈哈。它認(rèn)為我是一個(gè)人而且是很有臉的人！

NotACookieMonster
"absconder" lmao whatt

“潛逃者”，哈哈哈笑得我屁股痛，怎么了？

Elle Chapo
That’s hilarious. I’d wear AI label shades all day just to see the incredibly offensive things a computer says about the world around me.

太搞笑了。我會(huì)整天戴著人工智能標(biāo)簽?zāi)R，只是想看看計(jì)算機(jī)對(duì)我周圍世界里的人，會(huì)說(shuō)些什么令人難以置信的冒犯性的話。

Looks at wife "vegetarian, hipster, tennis ball fetching intern"

看著我老婆：“素食主義者、潮人、撿網(wǎng)球的實(shí)習(xí)生”

guaip
middle-aged man: a man who is roughly between 45 and 65 years old
I’m 36. Right in the feels…

中年男子：大約45歲到65歲之間的男子
我36歲了。感覺(jué)還算是對(duì)的…

firesurfer
Apparently I’m a psycholinguist, nonsmoker, and premmie. (literally a premature infant)
1 out of 3 right isn’t awful.

顯然，我是一個(gè)心理語(yǔ)言學(xué)家，不吸煙，也是一個(gè)早產(chǎn)兒。（字面上是過(guò)早出生的嬰兒）
三分之一是正確的，并不可怕。