Placeholder Image

字幕列表 影片播放

  • What is consciousness?

    什麼是意識?

  • Can an artificial machine really think?

    人造機器真的能思考嗎?

  • Does the mind just consist of neurons in the brain,

    心智只是由腦的神經元所組成

  • or is there some intangible spark at its core?

    或有無形的聰明才智在腦的核心嗎?

  • For many, these have been vital considerations

    對多數人而言, 這一直是未來人工智慧的主要考慮因素。

  • for the future of artificial intelligence.

    但英國電腦科學家 阿蘭•圖靈 決定不理會這諸多問題,

  • But British computer scientist Alan Turing decided to disregard all these questions

    而支持更簡單的一種:

  • in favor of a much simpler one:

    電腦能和人一樣交談嗎?

  • can a computer talk like a human?

    這問題後來發展出 一個測量人工智慧的理念,

  • This question led to an idea for measuring aritificial intelligence

    也就是眾所周知的「圖靈測試」。

  • that would famously come to be known as the Turing test.

    在 1950 年的一篇論文《電腦機器與智能》,

  • In the 1950 paper, "Computing Machinery and Intelligence,"

    圖靈提出了以下的測試。

  • Turing proposed the following game.

    一位審判員和不被看見的 被測試者做文字對話,

  • A human judge has a text conversation with unseen players

    並評估他們的回答。

  • and evaluates their responses.

    要通過測試, 電腦必須能取代被測試者之一,

  • To pass the test, a computer must be able to replace one of the players

    但不會明顯改變最後結果。

  • without substantially changing the results.

    換言之,此電腦會被認定是有智慧 ──

  • In other words, a computer would be considered intelligent

    若其對話不易與人類的區分時。

  • if its conversation couldn't be easily distinguished from a human's.

    圖靈預言在 2000 年之前 ,

  • Turing predicted that by the year 2000,

    具備 100 MB 記憶體的電腦 將能輕易通過圖靈測試,

  • machines with 100 megabytes of memory would be able to easily pass his test.

    但他可能言之過早了。

  • But he may have jumped the gun.

    即使目前配備更多記憶體的電腦,

  • Even though today's computers have far more memory than that,

    也很少成功,

  • few have succeeded,

    而那些通過測試的

  • and those that have done well

    著重在欺騙審判員的小技巧,

  • focused more on finding clever ways to fool judges

    而非運用強大的運算能力。

  • than using overwhelming computing power.

    雖然從未接受真正測試,

  • Though it was never subjected to a real test,

    第一個程式聲明它已成功的是 ELIZA,

  • the first program with some claim to success was called ELIZA.

    它僅以簡短的程式腳本,

  • With only a fairly short and simple script,

    模仿心理學家,成功迷惑許多人,

  • it managed to mislead many people by mimicking a psychologist,

    鼓勵他們說得更多些,

  • encouraging them to talk more

    同時也反問他們的問題。

  • and reflecting their own questions back at them.

    另一個早期程式腳本 PARRY 採取相反的方式,

  • Another early script PARRY took the opposite approach

    模仿偏狂型精神分裂症病患

  • by imitating a paranoid schizophrenic

    一直把話題轉回它預設的執著事物。

  • who kept steering the conversation back to his own preprogrammed obsessions.

    它們成功愚弄人們 突顯出這測驗的弱點,

  • Their success in fooling people highlighted one weakness of the test.

    人們通常把許多和聰明才智無關的事物 視為「智慧」。

  • Humans regularly attribute intelligence to a whole range of things

    儘管如此,年度競賽 例如 羅布納獎(Loebner Prize)

  • that are not actually intelligent.

    已使測試變得更正式了,

  • Nonetheless, annual competitions like the Loebner Prize,

    審判員已預先知道

  • have made the test more formal

    其中有些他們的對話者是機器。

  • with judges knowing ahead of time

    雖然品質已有改善,

  • that some of their conversation partners are machines.

    但許多聊天機器人的程式員 是採用類似 ELIZA 和 PARRY 的方法。

  • But while the quality has improved,

    1997 年的獲獎者 Catherine

  • many chatbot programmers have used similar strategies to ELIZA and PARRY.

    能夠進行驚人地專注且聰明的對話,

  • 1997's winner Catherine

    但多半只在審判員願意談 有關 比爾•克林頓 的事時。

  • could carry on amazingly focused and intelligent conversation,

    最近的獲獎者 尤金•古斯特曼 (Eugene Goostman)

  • but mostly if the judge wanted to talk about Bill Clinton.

    是個偽裝 13 歲烏克蘭男孩的機器人,

  • And the more recent winner Eugene Goostman

    審判員將它不合邏輯及蹩脚的文法

  • was given the persona of a 13-year-old Ukrainian boy,

    視為是語言及文化的障礙所致。

  • so judges interpreted its nonsequiturs and awkward grammar

    同時,另一程式例如 Cleverbot 則採用不同的方法,

  • as language and culture barriers.

    藉著統計分析大量真正對話的資料庫,

  • Meanwhile, other programs like Cleverbot have taken a different approach

    以決定最佳回應。

  • by statistically analyzing huge databases of real conversations

    有些還儲存先前對話的記憶,

  • to determine the best responses.

    以便長期改善。

  • Some also store memories of previous conversations

    雖然 Cleverbot 的每次回應 聽起來極像人類,

  • in order to improve over time.

    但它缺乏前後一致的個性,

  • But while Cleverbot's individual responses can sound incredibly human,

    而且無法應付嶄新的話題,

  • its lack of a consistent personality

    因而明顯露出馬腳。

  • and inability to deal with brand new topics

    在圖靈時代的人 怎能預料到今日的電腦

  • are a dead giveaway.

    能駕駛太空船、

  • Who in Turing's day could have predicted that today's computers

    執行精細手術、

  • would be able to pilot spacecraft,

    能解開龐大的方程式、

  • perform delicate surgeries,

    卻仍難以處理最基本的小交談呢?

  • and solve massive equations,

    人類語言 原來就是一種驚人複雜的現象,

  • but still struggle with the most basic small talk?

    甚至連最大的字典也無法記錄。

  • Human language turns out to be an amazingly complex phenomenon

    一個簡單的暫停就能難倒聊天機器人, 例如「嗯……」

  • that can't be captured by even the largest dictionary.

    或沒有正確解答的問題。

  • Chatbots can be baffled by simple pauses, like "umm..."

    一個簡單的對話句子,

  • or questions with no correct answer.

    例如:「我從冰箱拿出果汁給他,

  • And a simple conversational sentence,

    但忘了檢查保存期限。」

  • like, "I took the juice out of the fridge and gave it to him,

    這需要很多基本理解 及直覺語法分析。

  • but forgot to check the date,"

    事實証明模擬人類對話

  • requires a wealth of underlying knowledge and intuition to parse.

    不是只要增加記憶體 及數據處理能力而已,

  • It turns out that simulating a human conversation

    當離圖靈的目標越近時,

  • takes more than just increasing memory and processing power,

    我們終究要去處理 所有有關「意識」的大難題。

  • and as we get closer to Turing's goal,

    翻譯:Helen Lin

  • we may have to deal with all those big questions about consciousness after all.

What is consciousness?

什麼是意識?

字幕與單字

單字即點即查 點擊單字可以查詢單字解釋

B1 中級 中文 TED-Ed 圖靈 測試 對話 電腦 記憶體

【TED-Ed】圖靈測試。計算機能代替人類嗎?- Alex Gendler (【TED-Ed】The Turing test: Can a computer pass for a human? - Alex Gendler)

  • 830 72
    黃于珍 發佈於 2021 年 01 月 14 日
影片單字