一直對A.I.這個話題很感興趣，所以連我這種不怎樣喜歡發言的人也想說幾句，如有任何錯漏歡迎指點指點。
關於A.I. 的意思
如果大家有玩電腦遊戲的話，應該都會有跟電腦"AI"對戰的經歷，其實我覺得這個遊戲過程是在進行另一種形式的Turing Test(會上題及那種用來分別人和電腦的測試)，只是我們的目的是娛樂而非分別跟你對戰的那位是人或是電腦。我想帶出的是如果閣下對AI的定義不至於嚴謹得要各方面行為與人一樣，能應付衣食住行各方面，有情緒，其實所謂"AI"早已遍及我們的生活，而以上只是一例，更想指出的是這些"AI"其實未必如想像般神奇，尤其像我這個唸計算機出身的人，看著一堆程式碼變成一些被稱為有Intelligent的東東，感覺更加奇怪。關於AI的定義，可以參考一下Wikipedia,
"the study and design of intelligent agents" where an intelligent agent is a system that perceives its environment and takes actions that maximize its chances of success.
其實還是很空泛，因為隨意寫一個程式都會就enviroment (input) 去take action that maximize its chances of success (output)。
"Why do A.I. need to be like human being?"的問題
我認為以上問題最大的問題是本末倒置，問錯了方向。研究A.I.的主要目的不是要模仿人，而是以像人一樣的能力去解決各種生活、生產上的問題。再以上面電腦遊戲為例，遊戲A.I. 被創造的目的不是為了模仿人，而是讓玩者從遊戲互動中得到樂趣。不過，會不會有一日當A.I.技術日益精進後，真的能造出像人一樣的機器，有感情，有「創造力」，行為上與人無法分別？我不懂得回答。但如果有這一日，有一個「人」能擁有和你一樣的行為，而背後僅是一大堆程式碼，則必需反問自己憑甚麼認為自己擁有所謂「意識」？又或是如果你認同自己有「意識」，則必須回應那堆「程式碼」有沒有「意識」？
本會承傳先賢「海納百川，有容乃大，尊崇道德，樂發文章」的宗旨，將於中大賤標之巔招覽天下群雄，務以學術交流論劍切磋為己任。
(其實跟Melvin的文章沒啥關係，不過是自由聯/亂想)
回覆刪除古時，人按自己的形象造神；
今天，一如神按自己的形象造人的神話，人按自己的形象造機械僕人。
廢話講完，說些有點建設性的話。
之前肥幫有關外星人一講，以人作為智慧的基準。聽眾認為取樣不足，這樣定義實在主觀武斷。這種批評雖正確卻不公平。因為“人”是人所知的唯一一種智慧。在對其他可能性一無所知的情況下，只能從已知到未知地推敲。雖肯定不准確，但別無他法。
再說，即使世上有其他種類的智慧，例如蘑菇的智慧，雲的智慧，地球儀的智慧，也不一定能與人的智慧有重疊之處，建立聯繫亦未必可能，亦即表示我們可能永遠無法知道是否真的有這些智慧。
另外，所謂虛構一種“他世界的智慧”，其實也不過是人的智慧的局部，了不起的充其量是人少有探索的局部，並未跳出“人”的框框(亦是Ben在啟蒙一講中指出，人的思想最終仍會受自身語言系統有限的可能組合限制，並非什麽都能想的觀點)。
雖然老套，但也不得不提科幻小說中只有極少作品有效地處理這個問題。沒錯，又是2001太空漫遊及其續集(by Arthur Clarke)，和Solaris (by Stanislaw Lem)。其他作品，不是一高興就假設全宇宙的智慧生物都是人性化的(The Hitchhiker's Guide to the Galaxy，by Douglas Adams)，就是聲明劇情對白實非人類能了解，作者不過努力將至寫成人的語言(夜幕低垂，基地，by Isaac Asimov).
這樣看來，從人的觀點出發，以人為中心看來雖然自大，卻竟然是唯一能做的事。
同理，我們想得到能與我們溝通並服務我們的工具(其實就是要AI知道的一定比我們知道的要少，才能受控)，而不是完全不按人的牌理出牌的煤氣的智慧，蠶豆的智慧。因此談人工智慧，亦只能“人”地思考。不是“need to be like human beings”，而是“they are like human beings by default”。
"Why do A.I. need to be like human being?"
回覆刪除An AI similar to human is helpful for psychology research.
There are many ways to make an AI.
If an AI, using a particular algorithm, behave similar to human, then it is possible that the human mind is using that algorithm too.
--------------------------
It is possible that
"全宇宙的智慧生物都是人性化的"
This does not mean we are special,
this means we are ordinary.
Instead of saying there are many forms of intelligence i.e. "蘑菇的智慧，雲的智慧，地球儀的智慧"
I prefer to say there are many forms of culture.
I wonder if this leap is sensible.
Culture, it is another big topic.
What shapes the human culture?
What could E.T. culture like?
How could two culture communicate and comprehend each other?
My personal though is that if the E.T. species
1. have sexes
2. is mortal
than we have a good chance to be able to understand each other.
Iknow nothing about algoithms, but personally I can hardly imagine that behaviours/mindsets and algorithms are not one-one mapping.
回覆刪除--------------------------
No I am not talking about cultures. I am really talking about "intelligence".
What does it mean? The three laws of thought are (at least seems to be) the foundations of human intelligence.
1. Rule of Identity 同一律 A=A
2. Rule of Contradiction 矛盾律 A =/= ~A
3. Rule of Excluded Middle 排中律 (A or ~A) is true
When talking about "other intelligence", I am talking about "intelligence" that do not work like this. Though hardly imaginable, at least I WISH such thing exists.
--------------------------
I think it is really good point that if E.T.
1. have sexes
2. is mortal
then it is possible to have mutual understanding.
This point is good because it is about biology, so fundamental that we always do not have it in mind while concentrating on philosophical bullshit. It reminds us the most important fact that human beings are biological entities whose only aim of living is reproduction, simply because they will die sooner or later.
But I would like to ask why you mention sex. Is it because of interpersonal relationship or pleasure or something else? And maybe we should restrict the condition to be "have two sexes".
--------------------------
A sudden though: what if the E.T. do not have the concept of "communication"?