如果有一天「人造人」寫出了小說｜麥克尤恩演講全文

2018年“21大學生國際文學盛典”10月26日在中國人民大學舉行，伊恩·麥克尤恩是今年的致敬作家。70歲的他因為這場活動第一次踏足中國：“這項活動終於把我和我太太帶到了你們這片偉大非凡的國土上，我從心底裡感謝你們。”

在活動現場，麥克尤恩發表了題為《如果有一天“人造人”寫出了小說》的演講，主題是“數字革命”。他由互聯網談到人工智能，再談到人造人。

作為小說家，麥克尤恩關心的是，如果人工智能發展到第一個人造人寫出了第一部有意義的原創小說時，我們該如何理解他們。而小說，將是我們藉以理解他們的最佳途徑。

演講者

伊恩·麥克尤恩

1 | 數字革命

一場現代性的意識轉變

我想要開啟一段短暫的路程，踏入不可知的未來。

我的出發點是在我們的有生之年已經發生的一項深刻的改變，而它影響的是這顆星球上的絕大多數成人，還有孩子。當然，我所說的就是數字革命。

今天我們尚處在這場革命的初級階段。也許歷史剛剛完成了第一章。接下來的章節會更加深刻地影響我們如何理解我們自身的人性，進而影響我們的文學和我們所有的藝術形式。此時此刻，這些新章節正在書寫之中。

我這一代人，出生於20世紀中葉，成長於模擬信號的世界中。我們寄信，我們在公共電話亭裡通話（信號很差）；要想了解世界信息，我們會伸手求助書架上的百科全書。我們得到的新聞永遠遲了一天。接著，我們不得不笨拙地過渡到一個數字宇宙中，為了應付各種數字任務，我們時常得求助於自己的子女，然後是我們的孫輩。

對於80後以及更年輕的人而言，他們成人之時，英特網業已成為了生活的一部分，那些被他們隨身揣在口袋裡的強大電腦不僅僅是有用的工具——它們已經成為了自我的延展。英特網幾乎已成為一個包圍、影響意識本身的巨大精神體。

我們學會了像在自己的腦海中漫遊一樣漫遊於網絡空間之中。我們與朋友，與廣闊的信息世界的連接速度同步於我們思維的速度。英特網成為了我們的存儲空間，成為了雄心、知識、關係、夢想與渴望的中心。對於我們大多數人而言，從學生到總統，沒有了互聯網，工作——甚至是生活——都是不可想象的。它以它最美好的面孔和最醜惡的形態囊括了人性。它囊括了我們。一旦由於某種技術原因丟失了網絡連接，我們立刻就會感到孤獨，感到失落。這種奇怪的感覺完完全全是現代性的。它代表了人類意識的一次轉變。

但這僅僅是開始。

2 | 人工智能

變革終將到來

過去十年間，我們目睹了計算機科學的一場革命。人工智能時代已經降臨。

25年前，一臺計算機打敗了一位國際象棋大師。那臺計算機的程序中塞滿了數千場象棋賽。每走一步棋，它都會演算出每一種可能性。但就在去年，另一臺計算機僅僅被輸入了比賽規則和要求取勝的指令。

除此之外，別無其它。

比賽開始了，它下出了一步又一步非同尋常、步步見血的妙招，而這些絕不是人類能夠想出的招數——譬如說，開局棄後。一臺機器再定義了人類的遊戲。

機器學習已經進入了第一個興旺期。利用算法，基於我們之前的購買選擇，人工智能已經能夠給我們提供書籍電影網上導購建議了。它還能夠規劃商業航班線路。它還將在自動駕駛設計中大放異彩。

| 與alphago對弈的柯潔

這一切會通向何方呢？許多個世紀來，在許多種不同的文化中，人們的腦海裡一直縈繞著一個夢。那個夢就是創造出一個人造版的我們。就像基督教的上帝用黏土造出第一個人那樣，如今我們自己或許有朝一日也能成為上帝，造出我們自己的第一個人造人。

一千年前，在歐洲大大小小的教堂裡，有的雕像會在某些特殊的日子裡，為人類的錯誤和罪孽而落淚。人們目瞪口呆地望著這一神蹟——一尊上了油彩的石像竟然活了過來！如今我們知道了——雕像裡面藏著一個水槽，裡面有一條金魚。金魚一遊動，水就會順著一根隱蔽的管道流到人像的眼睛裡。

19世紀初葉，一本小說誕生了，它成為了後來一切的文本基石。瑪麗·雪萊的《弗蘭肯斯坦》講述了一個與該書同名的科學家如何造出了一個人，用電賦予了他生命。弗蘭肯斯坦的造物最後成了一個殺人犯。這個故事後來成為了一個強有力的隱喻，告誡我們科學創造最終可能讓我們玩火自焚。

我們生來就有一種將生命投射到無生命體上的本能。

我們的新表親或許一開始會成為孩子們的玩伴——目前市場上已經有了這類產品的初級版。機器人可以幫助照料老年人，在人口迅速老齡化的日本，這件事正在成為現實。科技已經造出了栩栩如生的皮膚、眼睛、頭髮；今天的機器人已經可以跳舞，甚至可以接球了——這件事並不像我們看來的那麼簡單。要設計出能夠準確無誤地理解，使用語言的軟件則是一項更加艱鉅的任務。而最最艱鉅的任務則是創造出所謂的通用智能。

但這一天會到來的。唯一的問題就是，何時到來。

3 | 無法抗拒的誘惑

當人造人擁有意識，意味著什麼

也許這一天到來得不會像我們許多人想象的那樣快。

人工智能讓我們瞭解到，大腦究竟有多麼神奇——一臺只佔一升空間的水冷三維生物計算機。它包含了大約1000億個神經元，每一個神經元平均產生7000條輸入與輸出信號。神經元之間的連接數量超出了我們的想象力。而這一切的耗能只有25瓦——相當於一個小燈泡的功率。我們目前的科技水平遠遠達不到這樣的微型化程度，更不用說還要同時避免元件過熱問題了。我們甚至還沒有找到儲存電能的高效途徑。你在網上看到的那種笨重的機器人通常都插著一根電源線。不接電源的話，它們蹦躂不了多久就得充電。

還有，我們很容易忘記一點：象棋和生活不一樣。象棋是一個封閉系統。比賽的歷史與現狀都確鑿無疑。比賽的結局也確鑿無疑。而生活卻是一個開放系統——在各個層面上都不可預測。語言也是一個開放系統。要理解一個句子，我們必須調用關於外部世界的先驗知識。要理解詞義，語境是至關重要的。

但無論如何，那一天會到來的，人工智能會出現在筆記本、臺式機與大型計算機中——它們還會幫助人類設計先進的人造人——因為我們無法抗拒那個計劃的誘惑，那個古老的夢想，儘管依照我們自己的形象製造一臺機器或許並沒有太多的科學價值。但我們也可以說，孤零零的一臺筆記本，一臺沒有軀體的計算機永遠也沒法同我們建立起一絲一毫的相似之處，它們既無法瞭解理解我們，我們也無法理解它們。

隨著硬件與軟件的進步，我們將面對一個古老的問題。當一個人形物帶著善解人意的表情 ，溫暖的聲音，有智慧、有見識的舉止出現在你面前，而你知道這個造物就是在北京附近的一座工廠中生產出來的時候，這究竟意味著什麼呢？你的新朋友真的和你一樣有意識嗎？還是說，那只是他的設計帶給人的錯覺？他真的有自我，有悲喜，會懷戀過去，期待未來嗎？

這個問題有一個很粗淺的應對方法。偉大的計算機科學家艾倫·圖靈在1930年代認真思考過機器智慧的問題，我們不妨採納他提出的理念：如果你根本沒法判定一臺機器是否有意識，那你不妨就假定他有意識。畢竟，我們所有人類都必須假定彼此有意識，可我們從來得不到這一點的最終證明。還有一個方法可以應對機器意識的難題——但我想把這一點留到最後。

歸根結底，思維所依附的人腦是由物質構成的。而物質是由你的大腦和人造人的“大腦”所共享的。近100年來，理論物理學家一直在告訴我們，物質比常識所以為的要奇怪的多。

當我們最終發展到能夠接受生物大腦相對於無機物大腦並不享有任何特殊地位或特權時，我們立刻就會面對一系列有趣的問題。我們應該授予人造人以公民權和公民義務嗎？買賣或擁有這樣一個造物是否是不道德的，就像過去買賣或擁有奴隸是不道德的一樣？摧毀這樣一個造物是否構成謀殺？他們會不會變得比我們更聰明，搶走我們的工作？在我們今日的工廠裡，聰明但沒有思維的機器已經開始替代工人了。下一個也許就輪到醫生和律師了。接下來就是那個終極問題：人造人會征服我們，甚至是取代我們嗎？

這些正是科幻小說多年來一直在探索的問題。現在，這些問題終於到了需要回答的時刻了。

4 | 偉大的冒險

用小說嘗試預演未來

未來已然降臨。我們可以賦予一臺計算機怎樣的道德準則呢？

自主、自動駕駛車輛的生產商們已經開始面對這件事了。你的新車應該忠於誰呢？一個孩子突然躥上馬路，正好躥入了你的行駛路線。如果你猛打方向盤，一頭撞上一輛迎面駛來的大貨車，你就可以挽救孩子的生命。

這個選擇必須在須臾之間做出。大腦運轉遲緩的人類不太擅長快速釐清這類問題。你新買的自動駕駛汽車可以遵照編定的程序，將你的生命安全置於其他所有人的生命安全之上。或者，它也可以奉利他主義和社會公益之名，準備好了犧牲你的生命。這是一個我們在設計汽車自動駕駛軟件時不得不面對的道德選擇。

對於我們許多人而言，我們的童年都會經歷這樣一個階段——我們終於意識到了一個非常深奧的道理：這顆星球上生活著數十億人類，其中的每一個個體都有著和你的內心一樣鮮活、真實、不言自明的精神生活。也許，這就是道德觀在一個人的童年誕生的時刻，那一刻你開始理解每一個旁人對於他們自己而言也是真實的，就如同你對於你自己而言是真實的一樣。你希望別人如何待你，你就得如何待人 ；你還得試圖理解一個不同於你的他人究竟意味著什麼。

這就把我的話題引到了另一項發明之上，一項古老的發明，不需要電池驅動，也無需高深的科技，但在道德上和審美上卻高度複雜，當它登峰造極之時，美得無以倫比。我說的是各種形式的小說。要想進入別人的思想，要想衡量不同人的思想彼此之間的關係，以及它們與容納它們的社會之間的關係，小說依然是我們最好的途徑，最好的工具。小說家是在他人的思想之海上揚帆的水手。電影直觀易懂，也很引人入勝，但它並沒有像許多人預言過的那樣讓小說消亡。只有小說能呈現給我們流動在自我的隱秘內心中的思維與情感，那種通過他人的眼睛看世界的感覺。

如果我們真的準備好了——也許就在本世紀——創造出全新的有意識體，而他們的思想會漸漸踏上一條和我們截然不同的道路，那麼小說就將是我們藉以理解他們的最佳途徑。我將我的一生都獻給了這種藝術形式，我確信它可以進入這顆星球上任何一個男人，女人和小孩的頭腦中。因此，它也可以進入一個類人機器人的頭腦中。小說可以嘗試著預演我們未來的主觀意識，包括那些我們所發明的頭腦的主觀意識。在我們爭論究竟應該給我們的造物注入何種道德體系的過程中，我們將不可避免地面對並闡明三個問題：我們是誰，我們是什麼，我們想要什麼。而當一個人造人寫處了第一部有意義的原創小說時——如果真有這一天的話——我們將有機會通過我們所創造的這些“他者”的眼睛看見我們自己。這將確鑿無疑地證明一件事：一種全新的，有意識的造物已經降生在我們身邊了。一場偉大的冒險將就此展開，無論它帶來的會是美好還是恐怖。

謝謝。（演講翻譯稿由騰訊文化提供，段落有調整，小標題為編者所加）

—— 演講英文版原文 ——

I want to take a brief journey into the unknowable future. My starting point is the profound change that has happened in our lifetimes and has affected the majority of adults on this planet, and children too. I’m talking, of course, about the digital revolution. We are still in its early stages. Perhaps history has just concluded chapter one. The next chapters will have even greater consequences for how we understand our own humanity, and therefore it will have consequences for our literature and for all our arts. Those new chapters are being written now.

My generation, born in the mid twentieth century, grew up in an analogue world. We posted letters, we talked to each other from public phone boxes, (the reception was poor) and for information about the world we reached for encyclopaedias from off the shelves. We got our news a day late. Then we had to make an awkward transition to the digital universe and we often turned to our children and then our grandchildren for help with certain digital tasks.

For those born in the 1980s and since, who came of age when the internet was already a fact of life, the powerful computers that sit in their pockets are not merely useful devices – they have become an extension of the self. The internet has come to resemble a vast mind which envelops and influences consciousness itself. We have learned to move around cyberspace as we might move around in our own thoughts. We connect with friends as well as the wider world of information with the speed of thought. The internet has become our memory store, the locus of aspiration, knowledge, relationships, dreaming and yearning. For most of us, from student to president, work, even life, is impossible without the internet. In its best and worst aspects, it encompasses human nature. It contains us. When we lose our connection to the internet for some technical reason, we feel isolated, bereft. This strange sensation is thoroughly modern. It represents a shift in human consciousness.

But it’s only the beginning. The past ten years have seen a revolution in computer science. The age of artificial intelligence is now upon us. Twenty five years ago, a computer beat a grandmaster at chess. It was programmed with thousands of chess games. At every move, it rehearsed every possibility. But last year, a computer was given only the rules of the game and told to win. Nothing else.When it played, it made extraordinary and successfulmoves that no human had ever thought of, like an early sacrifice of the queen. A machine re-imagined the human game. Machine-learning had arrived at its first blossoming. Using algorithms, artificial intelligence is already advising us which books or movies to buy on-line, based on our previous choices.It plans the routes of commercial aircraft. It will play its part in the design of autonomous vehicles.

Where might this lead us? There is a dream that has haunted many different cultures for centuries. It is a dream of making an artificial version of ourselves. Just as the Christian god was once supposed to have made the first man out of clay, so we confront now the possibility of becoming like gods ourselves by making our own first artificial human. Nearly a thousand years ago there were in the churches and great cathedrals of Europe, statues that wept on certain special days at the errors and sins of mankind. People stood amazed at this wonder - a painted stone figure coming to life! Now we know - concealed inside the statue was a tank of water with a goldfish. As it swam about, it pushed water along a hidden tube to the figure’s eyes.

In the early nineteenth century a novel appeared that became a foundational text. Mary Shelley’s Frankenstein tells the story of a scientist of that name who builds a man which he brings to life with electricity. Frankenstein’s creature turns out to be a murderer. The tale has become a powerful metaphor for how the creations of science can turn against us.

We ourselves have an innate tendency to project life onto inanimate things. A small baby will smile at the crudest drawing of a face. Who has not wanted to kick the car when it won’t start? As a philosopher said to me once, we are capable of forming an emotional relationship with a fridge. But now we see in our future the possibility of making plausible, intelligent humanoids of the sort that our fiction and movies have been playing with for many years. Our new cousins might first be playmates for children – already there are some simple versions on the market. Robots can help look after elderly people, as is already happening in Japan, where the population is ageing fast. Technology has produced life-like skin, eyes, and hair; there are now robots who can dance and – not so easy as it seems to us – even catch a ball. It is much harder to devise software that can understand and use language without mistakes. Hardest of all is creating what’s known as general intelligence. But it will happen. The only question is when.

It might not be as soon as many think. Artificial intelligence has taught us just how wondrous the brain is - a one litre, liquid-cooled, three-dimensional biological computer. It containsabout a hundred billion neurons, with an average of seven thousand inputs and outputs for each neuron. The number of connections between the neurons is beyond our imagining. And it all runs on twenty five watts – the power of a dim lightbulb. There is nothing in our current technology that comes anywhere near this degree of miniaturisation, and all achieved without overheating. We don’t even yet have an efficient way of storing electricity. The clumsy robots you see on the internet are often attached to a power line. Either that, or they can’t last very long on their feet without recharging.

And it is easy to forget - chess is not the same as life. Chess is a closed system. There’s no doubt about the history or present state of the game. Nor is there any doubt about the end of the game. Life, on the other hand, is an open system – unpredictable at every level. Language too is an open system. To understand a sentence we have to bring prior knowledge of the outside world. Context is everything when it comes to understanding words.

But still, it will come, artificial intelligence in laptops, desktops and mainframe computers – and they will help in the design of advanced artificial humans – because we can’t resist the project, ourancient dream, even though it might not have much scientific value to make a machine in our own image. Though it could be said that a simple, isolated laptop, a computer without a body that resembles ours, can never be remotely like us or understand us, nor we understand them.

As the hardware and software advance, we will face an old question. What would it mean, to have this human shape in front of you with a sympathetic expression, a warm voice, an intelligent, well-informed manner and to know that this creature was manufactured in a factory not far from Beijing? Might your new friend actually be conscious like you? Or is he simply designed to give that impression? Does he have a self, can he feel sorrow and joy, nostalgia and anticipation?There’s a crude way of dealing with this question: to adapt the notion of the great computer scientist, Alan Turing, who was thinking hard about machine intelligence in the 1930s: if you can’t even tell whether a machine is conscious or not, you might as well assume that it is. After all, we all have to assume consciousness in each other. We can never have a final proof. There’s another way of dealing with the problem of machine consciousness – but I’ll keep that back until the very end.

Ultimately, the human brain, on which the mind depends, is made of matter. Matter is what your brain and the ‘brain’ of an artificial human will both share. Theoretical physicists have been telling us for almost a hundred years that matter is far stranger than common sense suggests. When we reach the point at which we accept that there is no special status or privilege in having a biological as opposed to an inorganic brain, then immediately we will confront a host of interesting problems. Should we grant the rights and responsibilities of a citizen to an artificial human? Will it be wrong to buy or own such a being, as people used to buy and own slaves? Will it be murder if we destroy such a being? Will they become cleverer than us, and take our jobs? Already, in our factories, clever but mindless machines are replacing workers. Doctors and lawyers could be next. Then the ultimate question: will artificial humans dominate us, or even replace us?

These are questions that science fiction has been dealing with for many years. Now, at last, the questions need answering. The future has arrived. What kind of moral principles might we grant to a computer? The manufacturers of autonomous, self-driving vehicles are already confronting this. What are the loyalties of your new car? A child runs out into the road, right into your path. You could save its life if you swerved and crashed head-on into a lorry coming towards you. This choice must be made in a fraction of a second. Humans, with their slow-moving brains, are not very good at thinking this through quickly. Your new autonomous car might be programmed to care for your life above all other’s. Or it might be ready, in the name of altruism and the social good, to sacrifice you. This is a moral choice we will have to make in the design of the automobile software.

For many of us, there comes a point in childhood when we become aware of a very difficult idea: Of the billions who live on this planet, every single individual has a mental life that is just as vivid and real and self-evident as your own. Perhaps in childhood this is the moment when morality begins, when you start to understand that everyone else is as real to themselves as you are to yourself. You must treat others as you would wish to be treated and try to understand what it is to be someone other than yourself.

This notion brings me to another invention, an older one that runs without batteries and is technically simple but morally and aesthetically highly complex and, at its best, extremely beautiful. I’m thinking of the novel in all its forms. It remains our best means, our best device for entering the minds of other people, and measuring the relation of minds to each other, and to the society in which they live. The novelist is a voyager on an ocean of other minds. Cinema is immediately accessible and compelling, but it has not driven the novel into extinction as many once predicted. Only the novel can give us the flow of thought and feeling within the privacy of selfhood, that sense of seeing the world through the eyes of others.

If we are poised, perhaps in this century, to create new kinds of conscious beings whose minds might begin to be diverge from our own, the novel will be one of our best means of understanding them. I’ve given my life to this form and I’m certain that it can enter the mind of any man, woman or child on this planet. It can therefore enter the mind of a humanoid robot. The novel can attempt to rehearse our future subjectivity, including the subjectivity of those whose minds we will invent. As we debate what kinds of moral systems we want to install in our creations, we will inevitably have to confront and define who and what we are, what we want. And when,or if,an artificial human ever writes the first original and meaningful novel, we will have the opportunity to see ourselves through the eyes of others who we ourselves created. This will be final confirmationthat a new kind of conscious being is among us. A great adventure, benign or horrific, will begin.

Thank you.

閱讀更多 新京報書評週刊 的文章

關鍵字: 小說 機器人 人工智能