放心，人工智能不會很快奪走你的編程飯碗

{"type":"doc","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"以前，我曾擔心人工智能會奪走我的工作，讓我失業。毫無疑問，每年有關人工智能的恐懼散播的言論越來越多。但是我寫代碼的時間越長，對機器學習的瞭解越多，我就越不擔心這個問題。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"最近，OpenAI 和微軟（擁有 GitHub）發佈了"},{"type":"link","attrs":{"href":"https:\/\/copilot.github.com\/","title":"","type":null},"content":[{"type":"text","text":"Copilot"}]},{"type":"text","text":"，這是一個由人工智能驅動的結對編程“程序員”。儘管它給人們留下了深刻的印象，但是我們離能夠自己編寫代碼的人工智能還很遙遠。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"然而，這並不意味着沒有短期影響。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Sam Altman 曾經發過一條"},{"type":"link","attrs":{"href":"https:\/\/twitter.com\/sama\/status\/1404100794245214215","title":"","type":null},"content":[{"type":"text","text":"推文"}]},{"type":"text","text":"，說："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"預測：人工智能將使在電腦前進行的工作的價格比現實世界中發生的工作的價格下降得更快。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"和大多數人 （包括我）的期望相反，這會產生奇怪的效果。"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"對我們來說，幸運的是，軟件工程並不只是在電腦前打字。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"大多數寫人工智能的人都沒有寫過代碼"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"假如你編寫了一些中等複雜的軟件，開發者已過時的說法很快就會不攻自破。你知道“編寫代碼”對開發者來說並非真正的工作。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"要說“軟件工程師的工作就是編寫代碼”，就像說“作家的工作就是寫幾個段落”一樣。儘管這是事實，但卻忽略了一點：不是任何作家都能夠寫出《權利的遊戲》這樣的作品。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"編寫代碼是最簡單的部分。針對規模設計、滿足客戶需求和處理現有設計約束非常複雜。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"現在，軟件開發者是架構師、工程師和碼農。但是當我們走向未來，我想我們將會看到前者更多，後者更少。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"繁重的工作正在消失"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"像 Copilot 這樣由人工智能驅動的編碼工具在做什麼呢？智能生成樣板代碼。它們取消了我們價值最低的任務。例如編寫循環數據的代碼，創建單元測試，以及在 Stack Overflow 上尋找答案。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"身爲開發者，我們管理的是精力，而非時間。每個遊離的想法和谷歌搜索都會耗費一些精力。優秀的開發者希望通過研究具有挑戰性的問題，而不是寫簡單的代碼來獲取高額回報。Copilot 可以幫他們做到這一點。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"對於大多數開發者來說，這是好消息。除非你整天都在寫 CRUD 應用程序……"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"由於低價值任務日益枯竭，開發者變得更像技術項目經理，我認爲我們將在下一個十年中看到轉型。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"代碼價值將降低"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"正確生成的代碼可以讓有能力的開發者用更短的時間來構建更多的技術。這一價值的大部分將由公司獲得。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"軟件工程師的薪水不會降低，工作也不會增加。他們只是用更少的時間來完成更多的工作。當軟件開發成本下降時，需求會增加。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"blockquote","content":[{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"在經濟學中，當技術進步提高了使用資源的效率（減少任何一種使用所需的數量）時，Jevons 悖論（有時是 Jevons 效應）就會發生，但資源消耗的速度上升是因爲需求增加。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"——"},{"type":"link","attrs":{"href":"https:\/\/en.wikipedia.org\/wiki\/Jevons_paradox","title":"","type":null},"content":[{"type":"text","text":"Jevons 悖論"}]},{"type":"text","text":"，維基百科"}]}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們可以看到，很多公司都會使用相同的資源來解決更多的問題，或者出現了更多的技術公司。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"降低成本帶來新機會"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"你現在也許可以建立那些以前過於耗時或昂貴的副業項目了。假如建造某樣東西所需的時間較短，那麼邊際效益低的項目現在就變得可行了。以前過於昂貴的機會現在可以尋求了。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"也許你想開發一個聚合一些數據的簡單應用。只是由於金錢獎勵太低，不值得花 40 個小時來開發。但現在開發這一應用只需 20 個小時，你也許需要重新考慮一下。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"對公司來說也是如此。在他們現在所做的事情之外，他們不可能尋求更低價值的機會。作爲一個企業家和程序員，我喜歡這一點。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"人工智能驅動的編碼在未來可能帶來一些風險"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"現在，Copilot 生成了樣板代碼。它比你現有的代碼編輯器更加智能。但是它不會給任何人帶來風險。也就是說，我們必須假設工具會繼續改進。這樣的話，我們可以想象未來會出現以下問題。"}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"高級開發人才輸送的枯竭"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"我們都是新手，曾經只能編寫 CRUD 應用程序。但是當我們不再需要人來編寫這些簡單的應用時會發生什麼呢？如何讓新開發者學到足夠的知識，提升價值鏈？"}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"只有最優秀的開發者才能獲得收益"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"無論哪種職業，前 1% 的收入都高於其他人。只要頂尖的開發者能完成 10 倍的工作，就有可能減少對中等技能開發者的需求。"}]},{"type":"heading","attrs":{"align":null,"level":4},"content":[{"type":"text","text":"無代碼工具將使軟件變成黑盒子"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"如果我們不寫代碼了，又怎麼能理解人工智能編寫的代碼？我們是否習慣於使用我們不瞭解的技術？在某些時候，作爲一個行業和社會，我們可能會面臨這些長期問題。"}]},{"type":"heading","attrs":{"align":null,"level":3},"content":[{"type":"text","text":"最後的想法"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"這是我不擔心人工智能剝奪我工作的最大原因。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"假如人工智能能夠編寫軟件，那就完蛋了。大家都被淘汰了，不只是開發者。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Simon de la Rouviere 曾在"},{"type":"link","attrs":{"href":"https:\/\/twitter.com\/simondlr\/status\/1044605498727780355","title":"","type":null},"content":[{"type":"text","text":"推特"}]},{"type":"text","text":"上轉發了一篇文章《"},{"type":"link","attrs":{"href":"https:\/\/forge.medium.com\/yuval-noah-harari-21-lessons-21st-century-what-kids-need-to-learn-now-to-succeed-in-2050-1b72a3fb4bcf#---0-349","title":"","type":null},"content":[{"type":"text","text":"在 2050 年，要想成功，孩子們需要學習什麼"}]},{"type":"text","text":"》（"},{"type":"text","marks":[{"type":"italic"}],"text":"What Kids Need to Learn to Succeed in 2050"},{"type":"text","text":"）："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Naval 對此推文"},{"type":"link","attrs":{"href":"https:\/\/twitter.com\/naval\/status\/1044622979974410240","title":"","type":null},"content":[{"type":"text","text":"回覆"}]},{"type":"text","text":"道：（在 2050 年）“人工智能能夠比人類更好地編寫軟件”，這就像說一個人工智能可以寫小說，創造藝術，以及比人類更好的思考能力。一種能夠開發出更好的軟件的人工智能，是能夠管理世界的人工智能，它使整篇文章變得毫無意義。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"儘管媒體進行了大肆炒作，但我們也不必擔心人工智能會剝奪你的編程工作。編程主要是一種創造性的職業。而就目前的情況來看，神經網絡根本不可能有新的想法。另外，軟件工程師是解決問題的人，而且我認爲問題沒有盡頭。所以，儘管技術上的失業是不可避免的，但這並不會持續太久。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"作者介紹："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"Chris The Data Guy，全棧開發者，數據科學家。熟悉 Ruby、Python、React、AWS 等。對加密貨幣和經濟感興趣。"}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","marks":[{"type":"strong"}],"text":"原文鏈接："}]},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null}},{"type":"paragraph","attrs":{"indent":0,"number":0,"align":null,"origin":null},"content":[{"type":"text","text":"https:\/\/towardsdatascience.com\/artificial-intelligence-is-not-taking-your-coding-job-anytime-soon-5bf0df677ee8"}]}]}