The Mirror
المرآة
镜子
Wayback, Part 5. Zürich, 2007 — a Swiss engineer building AI a decade early, and the line I’ve never forgotten: “We are still ten years ahead.”
واي باك، الجزء الخامس. زيورخ، ٢٠٠٧ — مهندسٌ سويسريٌّ يبني الذكاء الاصطناعي قبل عقد، وجملةٌ لم أنسَها: «نحن متقدّمون بعشر سنوات».
回溯,第五部。苏黎世,2007——一位瑞士工程师提前十年打造人工智能,以及我从未忘记的一句话:“我们领先十年。”
By 2007 I had watched the future arrive early enough times to trust the feeling in my chest. What I hadn't done yet was meet someone who'd been living inside that feeling even longer than I had.
That year took me to Zürich — to a small company out by the airport, in a town called Glattbrugg, with a name that told you exactly how early it was: semantic system ag. Its founder was a Swiss engineer named Walter Diggelmann. (For years afterward I misremembered his name as "Dingelman" — memory sands the edges off even the people who mattered most.) He was doing, quietly, in 2007, the two things the whole world would be shouting about twenty years later: artificial intelligence, and data mining. Neural networks. He even had a phrase for the frontier he worked — "bio computing" — and projects that read, now, like premonitions: systems built to find the patterns no human had the time to find, for clients as serious as a national police force.
I wasn't the scientist in that room. My job was the bridge — between Walter's research, the university where I was finishing my master's, and the people who might fund what he was building. I was there to help the world catch up to him. It was the role I kept ending up in my whole life: standing between a thing that was ready and a world that wasn't.
And one day Walter said to me, without any drama, the line I have never been able to forget: "We are still ten years ahead."
Ten years. In 2007. Do that small piece of arithmetic and you land in the late 2010s — exactly when the thing he was describing stopped being a Swiss engineer's conviction and became the thing you now hold a conversation with on your phone.
Here is what "semantic" meant, and why it was the whole game. Back then the web was still Web 1.0 — text a machine could store but never understand. Write "his face turned red" and the computer saw a colour. It had no idea the man was blushing — that blood had risen to his cheeks from embarrassment, not paint. It couldn't tell "he blushed" from "he painted his face red." It read the letters and missed the meaning. What Walter was chasing — what the world later filed under "the semantic web" — was teaching the machine to read meaning, not just marks on a screen.
That idea is the quiet ancestor of everything people marvel at today. Before any model could write your contract or your poem, something first had to teach machines to comprehend what humans write at all. The line runs straight from that little office in Glattbrugg — through the breakthroughs a decade later that finally taught machines to read a word in the context of the sentence around it — to the assistant in your pocket. Walter was right. The idea was sound. It was only ever waiting on two things: the meaning, which he was building by hand, and the raw computing power to run it, which arrived only in the last few years.
I think about that office more than I expected to — not for the technology, but for the man. Certain, unhurried, working a full decade early in a quiet town, and right. He was a mirror. I had spent my life feeling early and a little foolish for it — the boy paying thousands to dial London, the CTO handing out email addresses to people who wanted a phone number. And here was someone who had built an entire quiet career out of that exact feeling, and been proven right by time.
You don't get many moments where the future looks back at you and simply nods. Zürich was one of mine. 🇨🇭 🇯🇴
■
بحلول عام ألفين وسبعة كنتُ قد رأيتُ المستقبل يصل مبكرًا مرّاتٍ كافيةً لأثق بالإحساس. ما لم أفعله بعد هو أن ألتقي بمن عاش داخل ذلك الإحساس مدّةً أطول منّي.
ذلك العام أخذني إلى زيورخ، إلى شركةٍ صغيرةٍ قرب المطار، باسمٍ يُنبئك كم كانت مبكرة. مؤسّسها مهندسٌ سويسريٌّ اسمه فالتر ديغلمان. كان يفعل، بهدوء، في عام ألفين وسبعة، الأمرين اللذين سيصرخ بهما العالم بعد عشرين عامًا: الذكاء الاصطناعي، والتنقيبَ في البيانات.
وذاتَ يومٍ قال لي، بلا مبالغة، جملةً لم أنسَها قط: "نحن متقدّمون بعشر سنوات." عشرُ سنوات، في عام ألفين وسبعة. احسبها، لتصل تمامًا إلى اللحظة التي صار فيها الذكاء الاصطناعي ما تحادثه اليوم على هاتفك.
ما الذي كان يسعى إليه؟ آنذاك كانت الشبكة نصًّا تخزّنه الآلة ولا تفهمه. اكتُب "احمرّ وجهه"، فترى الآلة لونًا، لا تدري أنّ الرجل قد خجل. أن تُعلّم الآلة أن تقرأ المعنى، لا الحروف فقط، ذلك هو السلفُ الهادئ لكلّ ما يذهل الناس اليوم. كان فالتر مُحقًّا؛ الفكرة كانت سليمة، وكانت تنتظر فقط المعنى الذي كان يبنيه، والقدرةَ الحاسوبية التي لم تصل إلا مؤخّرًا.
أفكّر في ذلك المكتب أكثر ممّا توقّعت. كان مرآة. قضيتُ حياتي أشعر أنّي مبكرٌ وأحمقُ قليلًا، وها هو رجلٌ بنى مهنةً كاملةً من ذلك الإحساس نفسه، وكان على حقّ.
■
到二〇〇七年,我已经见过太多次未来提前到来,足以相信那种直觉。我还没做过的,是遇见一个比我更久地活在那种直觉里的人。
那一年,我来到苏黎世,机场附近一家小公司,名字就告诉你它有多超前。创始人是一位瑞士工程师,瓦尔特·迪格曼。早在二〇〇七年,他就在安静地做着二十年后全世界会大声争论的两件事:人工智能,和数据挖掘。
有一天,他毫不夸张地对我说了一句我从未忘记的话:"我们领先十年。"十年,在二〇〇七年。算一算,正好落在人工智能终于变成你今天在手机上与之对话的那一刻。
他在追求什么?那时的网络,是机器能存储却无法理解的文字。写下"他涨红了脸",计算机看到的是一种颜色——它不知道那个人是害羞了。教机器读懂含义,而不只是字母,这正是今天让人们惊叹的一切背后那位安静的祖先。瓦尔特是对的。这个想法是成立的,它只是在等待含义——那正是他在构建的——以及算力,而算力直到近几年才到来。
我想起那间办公室的次数,超过我的预料。他是一面镜子。我一生都觉得自己太早、而且有点傻——而这里有一个人,用同样的感觉建起了整整一番事业,并且被证明是对的。
■
Note: Wayback is a personal series — lived memory, told as memory. Where a company, system or institution can be checked, it can be found in the public record. Written in a personal capacity.
Join the discussion
Thoughtful disagreement welcome. Comments are moderated.
Loading…