IV. TECHNOLOGY SHOULD AMPLIFY THE BEST OF TECHNOLOGY AND THE BEST OF HUMANITY
IV. 科技應該放大科技的優點和人性的優點。

One of the hallmarks of poorly designed systems is that they force the human user to act like a machine in order to successfully complete a task . Machines shouldn’t act like humans—at least, not in the current design environment that lacks the framework to effectively integrate “humanness” with devices—and humans shouldn’t act like machines . Amplify the best of each, without expecting them to do each other’s jobs . “Affective Computing” (https://en .wikipedia. org/wiki/Affective_ computin ) involves studying and developing devices that can “recognize, interpret, process, and simulate human affects .” The term was first introduced in 1995 by Rosalind Picard, professor of media arts and sciences at MIT. We’ll say more about affective technology soon .
設計不良系統的一個特點是，它們迫使人類用戶像機器一樣行事才能成功完成任務。機器不應該像人類一樣行事，至少在目前缺乏有效整合「人性」與設備的設計環境中不應該如此，而人類也不應該像機器一樣行事。增強每個人的優勢，而不指望它們做對方的工作。 「情感計算」(https://en.wikipedia.org/wiki/Affective_computing) 涉及研究和開發能夠「識別、解釋、處理和模擬人類情感」的設備。這個術語最早由麻省理工學院媒體藝術與科學教授羅莎琳德·皮卡德於 1995 年首次提出。我們很快會更多地談論情感科技。

Think of the automatic faucet that turns on the water for you, but requires you to hold your hands in a very narrowly defined location during the entire process of washing—something few humans naturally do
想像一下自動水龍頭，它會為你打開水，但在整個洗手過程中需要你將雙手放在非常狹窄的位置，這是很少有人自然會做的事情

The best technology, on the other hand, amplifies the best parts of both machines and people. It never crosses their roles, or forgets who is who . All tech is designed by people at some point The responsibility is on us to make it not just more efficient, but more accepting of the humanness of its users . A person’s primary task should not be computing; it should be being human.
最好的科技，另一方面，放大了機器和人的最好部分。它從不越界，也不會忘記誰是誰。所有科技最終都是由人設計的。責任在我們身上，讓科技不僅更有效率，而且更能接納使用者的人性。一個人的主要任務不應該是計算，而應該是做人。

Being human means seeking food, fun, and social connection. It means improving the local environment, participating in the community, meeting or finding friends and family, participating in rituals or festivals . It means finding, creating, and performing meaningful work; learning constantly; and developing skills . Humans are problem solvers, but we also feel pain, love and friendship, jealousy, fear, happiness and joy. We feel a sense of accomplishment when we achieve our goals . We study religion and history, and ache for belonging .
做人就是尋找食物、樂趣和社交。它意味著改善當地環境，參與社區，與朋友和家人見面或尋找他們，參與儀式或節日。它意味著尋找、創造和從事有意義的工作；不斷學習；並發展技能。人類是問題解決者，但我們也感受到痛苦、愛和友誼、嫉妒、恐懼、幸福和喜悅。當我們實現目標時，我們會感到成就感。我們研究宗教和歷史，並渴望歸屬感。

We are also the ones who create the next steps in a field—something technology, by itself, cannot do . A machine can run a piece of code and even evolve that code if a human programs it to do so, but the human specialty of jumping a layer of abstraction and coming up with an insight that changes the fundamental way we do things is something that is unpredictable
我們也是在某個領域創造下一步的人，這是技術本身無法做到的。機器可以執行一段程式碼，甚至在人類對其進行編程的情況下進化該程式碼，但人類跳躍抽象層並提出改變我們做事方式的洞察力，是無法預測的。

We have a history and a set of skills that we’ve developed in response to our culture and our environment. Humans understand context. Computers cannot understand context unless humans train them to . Originally the problem of teaching a machine to identify an object was thought to be a trivial task, but decades later it remains one of the most difficult problems in machine learning Humans are still the best at object recognition, and machines can take those human insights and index them in order to make them available to other humans
我們擁有一段歷史和一套技能，這些都是為了應對我們的文化和環境而發展出來的。人類能理解上下文。電腦無法理解上下文，除非人類訓練它們。最初，教導機器識別物體的問題被認為是一個微不足道的任務，但數十年後，它仍然是機器學習中最困難的問題之一。人類仍然是最擅長物體識別的，機器可以利用這些人類的洞察力並對其進行索引，以便讓其他人類使用。

No matter how much human knowledge is put into a computer, it will never have the same needs as a living organism . It won’t seek friendship or experience hunger, need to pee or clean itself It doesn’t care about its environment as long as it’s able to function Computers don’t form families, or hang out in groups
無論將多少人類知識輸入電腦，它永遠不會有與生物體相同的需求。它不會尋求友誼或體驗飢餓，也不需要上廁所或清潔自己。只要能運作，它不在乎周遭環境。電腦不會組成家庭，也不會聚在一起。

During the ’90s, my father worked on voice concatenation systems (combining individual prerecorded words to create meaning) for a large telecom company in the Midwest . His task was to build a digital directory assistance system that allowed people to call a number and have an automated voice respond to it . First he worked with voice talent to record hundreds of thousands of words and phrases . Then he worked with linguists to stitch the words together so that text could be read back in a smooth and human-like way
在 90 年代，我父親在中西部的一家大型電信公司工作，從事語音串接系統（將個別預先錄製的詞語組合起來以創造意義）的工作。他的任務是建立一個數位電話查詢系統，讓人們可以撥打一個號碼，然後由自動語音回應。一開始，他與語音專業人員合作錄製了數十萬個詞語和片語。然後，他與語言學家合作將這些詞語拼接在一起，以便文本可以以流暢且類似人類的方式被朗讀。

I used to sit at the dinner table and discuss artificial intelligence for hours with my dad He didn’t like the idea of artificial intelligence, and insisted on reading me bedtime stories from a pair of books called The Evolution of Consciousness by Robert E. Ornstein and Naturally Intelligent Systems by Maureen McHugh .
我過去常坐在餐桌旁和爸爸討論人工智慧好幾個小時。他不喜歡人工智慧的概念，堅持要給我讀睡前故事，內容來自羅伯特·E·奧恩斯坦的《意識的演化》和莫琳·麥休的《自然智能系統》這兩本書。

When my dad and I discussed voice recognition and automated systems, he would always point out how difficult these systems were to build: “Computers don’t have human forms. They don’t grow up,” he told me . “They don’t understand what it’s like to walk outside into the sun, or feel grass beneath their toes . They’re a brain without a body. Because of that, they don’t understand things the way that humans do . The best thing computers can do is to connect humans to one another.”
當我和我爸爸討論語音辨識和自動系統時，他總是指出這些系統建立起來有多困難：“電腦沒有人的形態。它們不會長大，”他告訴我。“它們不懂得走出戶外感受陽光，或感受腳趾下的草地。它們只是一個沒有身體的大腦。正因為如此，它們無法像人類那樣理解事物。電腦最擅長的事情就是將人類連接在一起。”

What this ultimately got me to realize was that no computer can understand humans as well as we can understand one another. Therefore, the best interfaces don’t connect us to technology; they connect us to other people. Google is indispensable not because it provides all the answers, but because it connects us to what others have discovered or written— and they have the answers . The Google interface itself is almost invisible . We don’t look at it; we look at the search results . Google Search doesn’t try to act human—it helps humans to find one another
這讓我最終意識到的是，沒有任何電腦能像我們彼此之間理解一樣理解人類。因此，最好的介面不是將我們連接到科技，而是將我們連接到其他人。Google 是不可或缺的，不是因為它提供所有答案，而是因為它將我們連接到其他人所發現或撰寫的內容，而他們有答案。Google 介面本身幾乎是看不見的。我們不看它；我們看搜索結果。Google 搜尋並不試圖表現得像人類，它幫助人類找到彼此。

Google Search is an example of a system that amplifies humanness and makes the best use of a machine . You can think of it as a switchboard connecting humans to humans, through a series of bots that index the majority of human digital knowledge . Without bots indexing data, we could never find anything Google doesn’t determine the best result for us, but it does give us a series of results we can choose from, prioritized by their importance to other humans . From a given list of results, we can then understand which ones best pertain to our problem The bots themselves only index human knowledge and help with the search results; they do not choose the result for us
Google 搜尋是一個能夠增強人性並充分利用機器的系統的例子。你可以把它想像成一個將人與人連接的交換機，透過一系列的機器人來索引大部分的人類數位知識。如果沒有機器人索引數據，我們將永遠找不到任何東西。Google 不會為我們確定最佳結果，但它會給我們一系列我們可以選擇的結果，按照它們對其他人的重要性排序。從給定的結果清單中，我們可以了解哪些最適合我們的問題。這些機器人只是索引人類知識並幫助搜尋結果；它們不會為我們選擇結果。

Mouse inventor Douglas Engelbart defined “augmenting human intellect” as the use of technology to increase capability people to be able to better approach complex problems and situations, to gain knowledge to suit specific needs, and to finally derive solutions to problems d The lesson for designers and engineers is to focus on optimizing your technology so that it amplifies the tasks that humans are better at that machines; tasks like curation, working with context, understanding, being flexible, and improvisation . A computer can’t truly ^[1]
滑鼠發明者道格拉斯·恩格爾巴特將「增強人類智慧」定義為利用科技來提升人們處理複雜問題和情況的能力，以獲取符合特定需求的知識，最終找到解決問題的方法。對設計師和工程師的教訓是要專注於優化科技，使其增強人類擅長而機器不擅長的任務，如整理、處理上下文、理解、靈活應變和即興創作。一台電腦無法真正

understand or curate, and once it’s been programmed, it’s relatively inflexible . The better a system supports humans to do these things, the better the result!
了解或策劃，一旦被程式化，就相對不靈活。系統越能支持人類做這些事情，結果就會越好！

These may seem like obvious differences, but they’re worth acknowledging explicitly as we decide how to design the interactions between these two very different intelligences
這些可能看起來是明顯的差異，但在我們決定如何設計這兩種非常不同智慧體之間的互動時，值得明確承認

• [1] Engelbart, Douglas . “Augmenting Human Intellect: A Conceptual Framework . ”SRI Summary Report AFOSR-3223, 1962. (http://www.dougengelbart.org/pubs/augment-3906, htm'i)
  Engelbart, Douglas.「擴增人類智慧：概念框架」。SRI 摘要報告 AFOSR-3223，1962 年。（http://www.dougengelbart.org/pubs/augment-3906.htm）