http://www.rrdnyyy.com/post/oiXHHofQZYAzsW5B?share=enable_share
https://backchannel.com/the-ai-takeover-is-coming-lets-embrace-it-d764d61f83a#.9fe8v6hpv
The AI Takeover Is Coming. Let’s Embrace?It.
Yes, millions of low-paying, low-skilled jobs are increasingly at risk. But there’s also much to gain from the coming AI revolution.
OnTuesday,the White House released achilling/'t??l??/reporton AI and the economy. It began by positing that “it is to be expected that machines will continue to reach and exceed human performance on more and more tasks,” anditwarned/w?rn/ofmassive job losses.
chilling/'t??l??/
adj. 寒冷的；冷漠的；使人恐懼的；令人寒心的；呱呱叫的（等于chillin）
n. 冷卻；寒冷
v. 冷卻（chill的ing形式）
warned/w?rn/of
vt. 警告，提醒；通知
vi. 發出警告，發出預告
Yet tocounter/'ka?nt?/this threat, the government makes a recommendation that may soundabsurd/?b's?d/: we have toincreaseinvestment in AI. The risk to productivity and the US’s competitive advantage is too high to do anything butdouble down on it.
counter/'ka?nt?/
vi. 逆向移動，對著干；反駁
absurd/?b's?d/:
adj. 荒謬的；可笑的
n. 荒誕；荒誕作品
double down on it
This approach not only makes sense, but also is theonlyapproach that makes sense.It’s easy?—?and justified?—?to worry about the millions of individual careers that something likeself-driving cars and truckswillretool/?ri'tul/, but we also havechasmsof need that machine learning could help fill. Our medical system is deeplyflawed/fl?d/; intelligent agents could spread affordable, high-quality healthcare to more people in more places. Our education infrastructure is notadequately/'?dikwitli/preparing students for thelooming/'l?m??/economicupheaval/?p'hivl/; here, too,AI systems couldchip inwhere teachers are spread too thin. We might gain energy independence by developing much smarter infrastructure, as Googlesubsidiarys?b's?d??ri/DeepMind did for its parent company’s power usage. The opportunities are too great to ignore.
retool/?ri'tul/
vt. 重組；重新裝備
vi. 更換工具；重新裝備；更換機械設備
chasmsof/'k?z?m/
n. 峽谷；裂口；分歧；深坑
flawed/fl?d/
adj. 有缺陷的；有瑕疵的；有裂紋的
adequately/'?dikwitli/
adv. 充分地；足夠地；適當地
looming/'l?m??/
adj. 隱隱約約的；正在逼近的
upheaval/?p'hivl/
n. 劇變；隆起；舉起
subsidiary /s?b's?d??ri/
adj. 附屬的；輔助的
n. 子公司；輔助者
More important, we have to think beyond narrow classes of threatened jobs, because today’s AI leaders—at Google and elsewhere—are alreadylaying the groundwork/'gra?nd'w?k/for an even moreambitious vision,the former pipe dream that is general artificial intelligence.
groundwork/'gra?nd'w?k/
n. 基礎；地基，根基
To visit the front lines of the great AI takeover is to observe machine learning systems routinely/r?'tinli/drubbing?dr?b??/humans in narrow,circumscribed/?s?k?m'sra?b/domains. This year, many of the most visiblecontestants/k?n't?st?nt/in AI’s face-off with humanity haveemergedfrom Google. In March, the world’s top Go playerweathered a humbling/'h?mbli?/defeat against DeepMind’s AlphaGo. Researchers at DeepMind also produced a system that canlip-readvideos with an accuracy thatleaves humans in the dust. A few weeks ago, Google computer scientists working with medical researchers reported an algorithm that candetect diabeticretinopathyin images of the eye as well as anophthalmologist/?ɑfθ?l'mɑl?d??st/can. It’s an early steptoward/t?rd/a goalmany companies are nowchasing: to assist doctors byautomating/'?t?met/the analysis of medical scans.
routinely/r?'tinli/例行公事地；老一套地
drubbing?dr?b??/
n. 毆打；徹底擊敗
circumscribed/?s?k?m’sra?b/
contestants/k?n't?st?nt/
n. 競爭者；爭辯者
face-off 對決
humbling/'h?mbli?/
adj. 令人羞辱的
v. 羞辱（humble的ing形式）；使…謙恭；使…卑賤
diabetic/?da??'b?t?k/
adj. 糖尿病的，患糖尿病的
n. 糖尿病患者
n. [眼科] 視網膜病
ophthalmologist/?ɑfθ?l’mɑl?d??st/
n. 眼科醫師
steptoward/t?rd/朝著
Also this fall, Microsoftunveiled??n'veild/a system that cantranscribe/tr?n'skra?b/human speech with greater accuracy than professionalstenographers/st?'nɑɡr?f?/. Speech recognition is the basis of systems like Cortana,Alexa, and Siri, and matching human performance in this task has been a goal for decades. For Microsoft chief speech scientist XD Huang, “It’s personally almost like a dream come true after 30 years.”
unveiled?/?n'veild/
adj. 裸露的；公布于眾的
v. 公開（unveil的過去分詞）；原形畢露
transcribe/tr?n'skra?b/
vt. 轉錄；抄寫
stenographers/st?'nɑɡr?f?/速記員
But AI’s 2016 victories over humans are just the beginning. Emerging research suggests we will soon move from theseslim/sl?m/slivers/'sliv?/of intelligence to something richer and more complex. Though a true general intelligence is at least decades away, society will still see massive change as these systems acquire anever-widening circle of mastery.That’s why the White House (well, at least while Obama’s still in office) isn’t shrinking from it.We are in the midst of developing a powerful force that will transformeverything we do.
slim/sl?m/
adj. 苗條的；修長的；微小的；差的
vt. 使…體重減輕；使…苗條
vi. 減輕體重；變細
slivers/'sliv?/
n. 裂片；條狀碎木片；廢屑（sliver的復數）
v. 成為薄片；裂成小片（sliver的第三人稱單數形式）
ever-widening不斷擴大的
To ignore this trend?—?to not plunge headlong into understanding it, shaping it, monitoring it?—?might well be the biggest mistake a country could make.
The tool of choicein theaforementionedexamples of successful AIs is deep learning: the artificial intelligence technique that’s beenrivaling/'ra?vl/habanerosinblistering'bl?st?r??/hotness/'h?tnis/.Its special nature is the reason we’re on thebrink/br??k/of a more general intelligence.
aforementioned/?,f?r’m?n??nd/
rival
n. 對手；競爭者
vt. 與…競爭；比得上某人
vi. 競爭
adj. 競爭的
blistering'bl?st?r??/
adj. 猛烈的；極熱的，極快的
n. [涂料] 起泡；發皰
v. 起水皰；起氣泡；使受暴曬（blister的ing形式）
adj. 上述的；前面提及的
hotness
n. 熱烈；熱心；暑熱
brink/br??k/
n. （峭壁的）邊緣
Though we’ve been able to train AIs to solve tasks for decades, experts had topainstakingly/?pens?tek? ?l?/hand-engineer manybespoke/b?'spok/components for every application. The years of human work needed to support an AI in recognizing objects in an image, for example, were totally useless/'jusl?s/for the problem ofdeciphering/di'saif?/sounds fortranscription/tr?n'skr?p??n/. In other words, we’ve had topre-chewour AIs’ food, over and over and over again.
painstakingly/?pens?tek? ?l?/
adv. 煞費苦心地；費力地
bespoke/b?’spok/
adj. 定做的；預定的
vt. 預約，顯示出
deciphering/di'saif?/
n. [通信] 解密
v. 破譯（decipher的ing形式）；解釋；辨認
transcription/tr?n'skr?p??n/.
n. 抄寫；抄本；謄寫
pre-chew/t??/咀嚼
The lesson of the past four years is that thistedious/'tid??s/pre-chewing is now, for the moment at least, largelyirrelevant/?'r?l?v?nt/. Instead, there’sessentially/?'s?n??li/one algorithm (with many minor variants) that can adjust its own structure to solve a problem, directly from whatever massively large data set you feed it. The result is not only better-performing systems, but also much fasterexperimentation/?k,sp?r?m?n'te??n/. “Many, many problems that welabored/?leb?d/onfor a long time and made very, veryhalting/'h?lt??/progress on, now in six months we can basicallyplow/pla?/throughthem,” says Google vice president and engineering fellow/'f?lo/Fernando Pereira.
tedious/‘tid??s/
adj. 沉悶的；冗長乏味的
irrelevant/?’r?l?v?nt/
adj. 不相干的；不切題的
experimentation/?k,sp?r?m?n'te??n/.
essentially/?'s?n??li/
adv. 本質上；本來
labored/?leb?d/on
n. 勞動；工作；勞工；分娩
vi. 勞動；努力；苦干
vt. 詳細分析；使厭煩
halting/'h?lt??/
adj. 猶豫的；蹣跚的；跛的
v. 停止；蹣跚；猶豫（halt的ing形式）
plow/pla?/through
vi. [農機] 犁；耕地；破浪前進；開路
vt. [農機] 犁；耕；開路
n. [農機] 犁；似犁的工具；北斗七星
Yet as impressive as human-quality speech recognition, lip reading and image tagging are, it’s not immediately obvious that they’re thecornerstones/'k?rn?ston/of some great, all-powerful intelligence. It’s somewhat like having your kid come home with areport cardof As in subjects that include English,knitting/'n?t??/the heels/hilz/of socks,dodgeball/'dɑd?'b?l/, and calculating ahypotenuse/ha?'pɑt?nus/. You’d likely wonder if this clever kid will be able to draw connections between those areas to emerge as a critical thinker. So is deep learning really on a path to challenging true human intelligence?
cornerstones/'k?rn?ston/
n. 基礎；柱石；地基
report card
成績單
工作報告
knitting/'n?t??/
heels/hilz/
n. 高跟鞋（heel的復數）；腳踝；殘余料
v. 緊跟；給（鞋等）裝跟（heel的三單形式）
dodgeball/‘dɑd?'b?l/
https://en.wikipedia.org/wiki/Dodgeball
n. 躲避球
hypotenuse/ha?'pɑt?nus/直角三角形的斜邊
“The reason we’re seeing extremely narrow systems right now is because they’re extremely useful,” says Ilya Sutskever, cofounder and research director of OpenAI. “Good translation is extremely useful. Good cancer screening is extremely useful.So that’s what people are going after.”
But he adds that although today’s systems look narrow, we “are already beginning to see the seed ofgenerality/?d??n?'r?l?ti/.” The reason is that the underlying techniques are all justmild/ma?ld/riffson one concept. “These ideas are so combinable, it’s likeclay/kle/. You mix and match them and they can all be made to work.”
generality/?d??n?'r?l?ti/.
n. 概論；普遍性；大部分
mild/ma?ld/
adj. 溫和的；輕微的；淡味的；文雅的；不含有害物質的的
n. （英國的一種）淡味麥芽啤酒
n. (Mild)人名；(瑞典)米爾德；(德、捷、芬)米爾德
riffs/r?f/
n. 反復樂節；即興重復段
n. (Riff)人名；(法、葡、匈)里夫
mild riffs溫和的段子
clay/kle/.
n. [土壤] 粘土；泥土；肉體；似黏土的東西
vt. 用黏土處理
n. (Clay)人名；(英、法、西、意、葡)克萊
combinable[k?m’bain?bl]
可以化合的
組合式
可以結合的
By mixing and matching the narrow systems of today, we’llland onsomething bigger and broader?—?and more recognizable as intelligent—tomorrow.
land on
登陸
降落
猛烈抨擊
著陸
One early, tantalizing/?t?ntl..a?z??/exampleof what higher intelligence might eventually look like comes from Google’s translation research. In September, Google announced anenormous/?'n?rm?s/upgrade in the performance of Google Translate, using a system it’s calling Google Neural Machine Translation (GNMT). Google’s Pereira called the jump in translationquality/'kwɑl?ti/“something I never thought I’d see in my working life.”
tantalizing/?t?ntl..a?z??/
adj. 撩人的；逗引性的；干著急的
v. 惹弄；逗弄人（tantalize的ing形式）
enormous/?'n?rm?s/
adj. 龐大的，巨大的；兇暴的，極惡的
quality/'kwɑl?ti/
n. 質量，[統計] 品質；特性；才能
adj. 優質的；高品質的；<英俚>棒極了
“We’d been making steady progress,” he added. “This is not steady progress. This isradical./'r?d?kl/”
radical./'r?d?kl/
adj. 激進的；根本的；徹底的
n. 基礎；激進分子；[物化] 原子團；[數] 根數
With the new Translate nowrolling outlanguage by language, some Googlers decided to go even further. They wondered if they could build a single translation system that couldjuggle/'d??ɡl/many languages and potentially displaytransfer learning, ahallmark/?h?l?mɑrk/of human intelligence. Transfer learning is the ability to apply one skill, such as playing the piano, to speed up theacquisition/??kw?'z???n/of another, such asconducting/k?n'd?kt/anorchestra/'?rk?str?/or learning another instrument.
rolling out
鋪開；滾出
juggle/'d??ɡl/
vi. 玩雜耍；欺騙；歪曲
vt. 歪曲；欺騙
n. 玩戲法；欺騙
盡力對付
hallmark/?h?l?mɑrk/
n. 特點；品質證明
vt. 給…蓋上品質證明印記；使具有…標志
n. (Hallmark)人名；(英)霍爾馬克
conducting/k?n'd?kt/
vi. 導電；帶領
vt. 管理；引導；表現
n. 進行；行為；實施
orchestra/‘?rk?str?/
n. 管弦樂隊；樂隊演奏處
It seems obvious to us that knowing the fundamentals of music would help apianist/'p??n?st/pick up theukulele/'j?k?'leli/, but that’s not how language translation has been done. In GNMT, one deep learning system had toabsorb/?b?s?rb/millions of German-to-English translations, and teach itself how to take inder rote Hundand spit outthe red dog. A separate system independently learned how to translate in the other direction, from English to German. Same goes for French to English, English to French, Korean to Japanese, and so on?—?every pair of languages uses its owndistinct/d?'st??kt/system, built as if the act of translation was being inventedaneweach time. To support translation between 100 languages, you might end up training almost 10,000 separate systems. That’s time consuming.
ukulele/'j?k?'leli/,尤克里里琴（夏威夷的四弦琴，等于ukelele）
absorb/?b?s?rb/
vt. 吸收；吸引；承受；理解；使…全神貫注
distinct/d?'st??kt/
adj. 明顯的；獨特的；清楚的；有區別的
These researchers wanted to know if they could build a single model for multiple languages that could hold its own against those one-off systems. First, it might be more efficient. And maybe something interesting would emerge from having all those words and languagesjangling/'d???ɡl/aroundinside a single architecture.
jangling/'d???ɡl/
n. 爭吵，吵嚷；刺耳聲
vt. 使發出刺耳聲；使爭論
vi. 刺耳響；爭論，吵架
They started small, with a neural network trained on Portuguese/?p?rt???ɡiz/and English, and on English and Spanish. So far so good: this single multilingual/?m?lt?'l??ɡw?l/system did almost as well as the state-of-the-art, dedicated GNMT models in translating between English and either Spanish or Portuguese. Then they wondered: could this algorithm also translate between Portuguese and Spanish?—?even though it hadn’t seen a single example of Portuguese-Spanish translation?
multilingual/?m?lt?'l??ɡw?l/
adj. 使用多種語言的
n. 使用多種語言的人
As theyreportedin November, the result they got was “reasonably good quality”?—?notstaggering/'st?ɡ?r??/in its perfection, but not bad for anewbie/'nubi/. But when they then fed it a small set of Portuguese-to-Spanish sentence pairs, sort of anamuse/?'mjuz/bouche/bu:?/ofdata, the system suddenly became just as good as a dedicated GNMT Portuguese-to-Spanish model. And it worked for other bundles of languages, too. As the Google authors write in the paper, this “is the first time to our knowledge that a form of true transfer learning has been shown to work for machine translation.”
staggering/'st?ɡ?r??/
adj. 驚人的，令人震驚的
stagger['st?ɡ?]
vt. 蹣跚；使交錯；使猶豫
vi. 蹣跚；猶豫
n. 蹣跚；交錯安排
adj. 交錯的；錯開的
newbie/'nubi/.
n. 網絡新手；新兵
amuse bouche
餐前點心，可口小吃
It’s easy to miss what makes this so unusual. This neural net had taught itself arudimentary/?rud?'m?ntri/new skill using indirect information. It had hardly studied Portuguese-to-Spanish translation, and yet here it was,acingthe job. Somewhere in the system’sguts/ɡ?ts/, the authors seemed to see signs of a sharedessence/'?sns/of words,a gist/d??st/ofmeaning.
rudimentary/?rud?'m?ntri/
adj. 基本的；初步的；退化的；殘遺的；未發展的
guts/ɡ?ts/
n. 內臟；飛碟游戲（比賽雙方每組5人，相距15碼，互相擲接飛碟）；狹道；貪食者（gut的復數）
v. 取出…的內臟；毀壞…的內部；貪婪地吃（gut的第三人稱單數）
n. (Guts)人名；(德)古茨
n. （俚語）勇氣；決心
essence/‘?sns/
n. 本質，實質；精華；香精
gist/d??st/
n. 主旨，要點；依據
Google’s Pereira explains it this way: “The model has a common layer that has to translate from anything to anything. That common layer represents a lot of the meaning of the text, independent of language,” he says. “It’s something we’ve never seen before.”
Of course, this algorithm’s reasoning power is very limited. It doesn’t know that a penguin is a bird, or that Paris is in France. But it’s a sign of what’s to come: an emerging intelligence that can makecognitive leaps/li:p/based on an incomplete set of examples. If deep learning hasn’t yet defeated you at a skill you care about, just wait. It will.
leaps/li:p/
vi. 跳，跳躍
n. 飛躍；跳躍
vt. 跳躍，跳過；使躍過
n. (Leap)人名；(法)萊亞
Training one systemto do many things is exactly what it takes to develop a general intelligence, andjuicing upthat process is now a core focus of AI boosters. Earlier this month OpenAI, the researchconsortium/k?n's?rt??m/dreamed upby Elon Musk and Sam Altman, unveiledUniverse, an environment for training systems that are not just accomplished at a single task, but that canhop/hɑp/around andbecomeadept atvarious activities.
juicing up
使…活躍；使…有精神；使…更動人
consortium/k?n's?rt??m/. 財團；聯合；合伙
hop/hɑp/around跳來跳去
v. 單足跳躍〔跳行〕
vt. 搭乘
vi. 雙足或齊足跳行
n. 蹦跳,跳躍；跳舞；一次飛行的距離
As cofounder Sustkever puts it, “If you try to look forward and see what it is exactly we mean by “intelligence,” it definitely involves not just solving one problem, but a large number of problems. But what does it mean for a general agent to be good, to be intelligent? These are not completely obvious questions.”
So he and his team designed Universe as a way to help others measure the general problem-solving abilities of AI agents. It includes about a thousand Atari games, Flash games, and browser tasks. If you were to enter whatever AI you’re building into the training ring that is Universe, it would beequipped withthe same tools a human uses to manipulate a computer: a screen on which to observe the action, and a virtual keyboard and mouse.
The intent is for an AI to learn how to navigate one Universe environment, such asWing Commander III, then apply that experience to quickly get up to speed in the next environment, which could be another game, such asWorld of Goo, or something as different as Wolfram Mathematica. A successful AI agent would display some transfer learning, with adegree/d?'ɡri/ofagility/??d??l?ti/andreasoning.
agility/??d??l?ti/
n. 敏捷；靈活；機敏
degree/d?'ɡri/
n. 程度，等級；度；學位；階層
This approach is not withoutprecedent/'pr?s?d?nt/. In 2013, DeepMindrevealed/r?'vil/a single deep learning-based algorithm that discovered, on its own, how to play six out of seven Atari games on which it was tested. For three of those games?—Breakout,Enduro, andPong—?itoutperformeda human expert player. Universe is a sort of scaled-up version of that DeepMind success story.
precedent/'pr?s?d?nt/.
n. 先例；前例
adj. 在前的；在先的
v. 透露（reveal的過去式）；顯示
As Universe grows, AItrainees/trei'ni:/can start learninginnumerable/?'n?m?r?bl/useful computer-related skills. After all, it is essentially aportal/'p?rtl/into the world of anycontemporary/k?n't?mp?r?ri/deskjockey/'d?ɑki/. Thediversity/da?'v?s?ti/of Universe environments might even allow AI agents to pick up somebroadworld knowledge that otherwise would betough/t?f/to collect.
trainees/trei'ni:/
n. [經管] 實習生；[勞經] 受訓人員（trainee的復數）；訓練中的動物
innumerable/?'n?m?r?bl/
adj. 無數的，數不清的
portal/'p?rtl/
n. 大門，入口
n. (Portal)人名；(法、西、葡)波塔爾；(英)波特爾
jockey/'d?ɑki/
vt. 駕駛；欺騙；移動
n. 操作工；駕駛員；賽馬的騎師
vi. 當賽馬的騎師；耍手段圖謀；搞欺騙
contemporary/k?n't?mp?r?ri/
n. 同時代的人；同時期的東西
adj. 當代的；同時代的；屬于同一時期的
tough/t?f/
adj. 艱苦的，困難的；堅強的，不屈不撓的；堅韌的，牢固的；強壯的，結實的
n. 惡棍
vt. 堅持；忍受，忍耐
adv. 強硬地，頑強地
n. (Tough)人名；(英)圖赫
It’s a bit of aleap/lip/from a Flash-and-Atarichampion/'t??mp??n/to an agent that improves the quality of healthcare, but that’s because our intelligent systems are still in kindergarten.For many years, AI hadn’t made it even this far. Now it is on the path to first grade, middle school, and eventually, advanced degrees.
leap/lip/
vi. 跳，跳躍
n. 飛躍；跳躍
vt. 跳躍，跳過；使躍過
n. (Leap)人名；(法)萊亞
champion/‘t??mp??n/
n. 冠軍；擁護者；戰士
vt. 支持；擁護
adj. 優勝的；第一流的
n. (Champion)人名；(英)錢皮恩；(法)尚皮翁
Yes, the outcome is uncertain. Yes, it’s totally scary. But we have a choice now.We can try to shut down thismurky/'m?ki/future that we can neither fully control nor predict, and run the risk that the technologyseeps/sip/outunbidden/?n'b?dn/,potentially/p?'t?n??li/triggering/'tr?g?/massivedisplacement/d?s'plesm?nt/. Or we can actively/'?ktivli/try to guide it to the paths of greatest social gain, and encourage/?n'k??d?/the future we want to see.
murky/'m?ki/future
adj. 黑暗的；朦朧的；陰郁的
seeps out
vi. 漏；滲出
n. 小泉；水陸兩用的吉普車
unbidden/?n'b?dn/
adj. 未受邀請的；未受指使的；自愿的
actively/'?ktivli/
adv. 積極地；活躍地
encourage/?n’k??d?/
vt. 鼓勵，慫恿；激勵；支持
I’m with the White House on this one. A deep learning-powered world is coming, and we might as well rush right into it.
Creative Art Direction:Redindhi Studio
Illustration by:Laurent Hrybyk
【time】
6:16 - 6:58 am 42m
6:58-7:12am14m
9:26-10：54pm1.29m
[sentence]
This approach not only makes sense, but also is the only approach that makes sense.
It’s personally almost like a dream come true after 30 years.
We are in the midst of developing a powerful force that will transform everything we do.
Its special nature is the reason we’re on the brink /br??k/ of a more general intelligence.
So that’s what people are going after.
something I never thought I’d see in my working life.
an environment for training systems that are not just accomplished at a single task, but that can hop/hɑp/ around and become adept at various activities.
Yes, the outcome is uncertain. Yes, it’s totally scary. But we have a choice now.
I’m with the White House on this one. A deep learning-powered world is coming, and we might as well rush right into it.
