林棟樑 | Don Lin
在與AI和機器人的競爭中,人類的失業率會高到九成九?
這一波的人工智慧(AI)革命不是像以前Goole Search那麼簡單. 以前Google Search,你問它來一段七言絕句, 他會從非常大的資料庫search, 但是搬出來的就是裡面蒐集有的詩句, 不會自動創造新的. 現在的AI, 你可以給它一張你當年大學畢業拿文憑與父母親合照的照片, 叫它寫一首押普通韻的七言絕句來表示畢業的興奮, 同時製作一段10秒鐘的video表示當時歡欣雀躍的心情. AI在不到一分鐘就會寫出頗有水準的七言絕句, 以及一段讓你看了都會熱淚盈眶的video. 這七言絕句及video, 是AI用它的generative (生成式) power巧妙的創作出來. 再者, AI已經會寫電腦軟體. 筆者就在2025年開發的天象圖的App時, 就找AI幫忙寫網站用的coding. AI寫coding的速度, 一分鐘不到, 幾百條的軟體程式就寫出來, 而且run起來, 居然通通過關, 不可思議.
除了AI外, 發展人形機器人來幫忙人類的各種勞動也是許多大科技公司的重點發展. 以前我們都看過robots在汽車製造工廠裡面幫忙組裝汽車, 現在的智能汽車, 像Tesla, 用robot來幫人開車, 這些都不是天方夜譚而且是實實在在發生在我們眼前的事.
這些高科技的成果之所以變成可能, 歸根究柢就是因為在1947年美國貝爾實驗室(Bell Labs)的物理學家發明了半導體, 可以快速控制二位元(binary)邏輯線路的運作. 速度決定一切. 從此人類文明轉了一個大方向, 也就是電腦的世界. 各式各樣的電腦在過去這50多年來在職場或家庭生活上幫了很大的忙. 但是像過去一樣, 人類的biological intelligence在方方面面俱有無限的創造力, 連artificial intelligence都能創造出來. 這AI越來越厲害, 結合機器人的發展, 會不會威脅到人類, 特別是在人類社會最重要的job market這一塊.
有的人是憂心忡忡. 像Louisville大學電腦教授Roman Yampolskiy就在最近的podcast說在2030年AI會讓99%的人失業. 即使是電腦程式設計師和某些AI工程師也無可避免地受到自動化浪潮的衝擊. 在人類社會把獲取最高利潤當做準則的大趨勢之下, 能用AI輕易取代的人工將被無情地淘汰. 這一波歷史上前所未有的失業巨浪襲來的速度之快, 連政府都沒辦法訂下任何對策. 而絕大部分以工作作為生命意義重心指標的勞工, 在失業浪潮打擊之下將會覺得生命沒有任何其他的意義, 因而引起的種種社會問題,將會對人類社會產生極大的衝擊. 他預測, 在種種領域中能力與人類相當的AI+robotics系統將在 2027 年實現. 經過3年的橫掃千軍, 勞動市場極有可能全面崩潰. 10%的市場失業率已經夠嚇人, 但是他講的是99%!
雖然沒有講得這麼嚇人, 但是華人中天王級人物, NVIDIA的CEO, 黃仁勳先生(Jensen Huang)也看出AI跟robotics的衝擊. 他在2024年3 月史丹佛經濟政策研究所 (SIEPR) 經濟高峰會的一場演說中預測說, 五年內, 人工智慧將通過所有人類測試（例如, 律師資格考試、行醫執照）, 也就是說, 教育、職業和知識型工作將發生重大轉變. 他又預測, 十年內, 人工智慧的運算能力將提升百萬倍, 令AI如虎添翼. 在2024 年8月的丹佛電腦圖形學和互動技術的年度會議SIGGRAPH中他也表示, 人工智慧將改變每個人的工作, 包括他自己的工作. 但他強調, 人工智慧是顯著地增強人類能力而非取代人類. 至於失業問題, 他在2025 年 7 月的CNN 採訪中有提到, 如果人類沒有新的想法, 人工智慧驅動的生產力可能會導致失業, 但回顧歷史, 科技不僅增加就業, 也增加了生產. 他稱人工智慧是 ”greatest technology equalizer”, 它讓非技術人員輕易取得本來他們不會的技能, 令眾多人的 ”豐富的創意” 都可以實現. Meta的Yann LeCun也持同樣樂觀的態度, 認為人工智慧將轉變工作, 而不是徹底取代它們．
其他, RethinkX的Adam Dorr預測, 到2045年將出現大規模失業. 被稱為「人工智慧教父」的Geoffrey Hinton則認為, 呼叫中心和律師助理等「平凡的腦力勞動」將被取代, 但他認為水管工等體力勞動相對安全. Anthropic執行長 Dario Amodei則表示, 一半的初級辦公室工作可能在五年內消失, 而OpenAI的Sam Altman則認為, 社會將適應並創造新的角色.
筆者在美國科技業工作4-50年, 而且最近也親自拜託Grok幫忙寫Javascript的網站軟體程式，所得的經驗是AI確實厲害得無話可說. 每次它給的近百條程式, 裏面邏輯密密麻麻. 然後去execute所給的程式, 得出的結果竟然分毫不差, 讓人整個下巴掉下來！一年多來用了7-8種AI, 並常常看其他人如何利用AI, 也覺得AI正在人類文明不同領域產生不可思議的變化. 但是個人還是覺得, AI在5年內會取代世界百分之九十九的jobs是太危言聳聽. 如果10年內會取代百分之三十的jobs應該是比較reasonable. Even so, 我們很難想像10年後職場能夠繼續survive的job到底是什麼樣子, 而在這麼高的失業率之下, 整個人類社會是否會變得「面目全非」. (2025-09-04 New Jersey)
Would AI and robots cause the unemployment rate to reach 99%?
Th wave of artificial intelligence (AI) revolution is not as simple as Google Search. Previously, if you asked Google Search for a seven-character quatrain, it would search a vast database, but it would only retrieve existing verses; it wouldn’t automatically create new ones. Today, you can feed AI a photo of you, graduating from college with your parents on the side and receiving your diploma, and ask it to write a seven-character quatrain in a standard rhyme to express your excitement, along with a 10-second video capturing your joyful feelings. In less than a minute, the AI will produce a highly competent seven-character quatrain and a tear-jerking video. These quatrains and videos are skillfully created by AI using its generative power. Furthermore, AI can write computer software. When I was developing a sky map type of app in 2025, I enlisted AI to help write the relevant codes to be used on the website. The AI offered hundreds of lines of software in less than a minute, and then they ran flawlessly. It’s incredible.
In addition to AI, the development of humanoid robots to assist humans in various tasks is also a key focus for many major tech companies. We’ve all seen robots assembling cars in auto factories before, and now smart cars like Tesla’s use robots to assist drivers. These aren’t just fantasies; they’re happening right before our eyes.
These high-tech achievements were ultimately made possible by the invention of semiconductors by physicists at Bell Labs in 1947, enabling the rapid operation of binary logic circuits. Speed is everything. From then on, human civilization shifted in a major direction, towards the world of computers. Over the past 50 years, various computers have greatly aided both workplaces and family life. However, as always, human biological intelligence possesses boundless creativity in all areas, even capable of producing artificial intelligence. With the increasing power of AI, combined with the development of robots, will this pose a threat to humanity, particularly in the crucial job market?
Some are deeply concerned. University of Louisville computer science professor Roman Yampolskiy, for example, stated in a recent podcast that AI will render 99% of the workforce unemployed by 2030. Even computer programmers and some AI engineers will inevitably be impacted by the wave of automation. As human society prioritizes maximizing profits, those easily replaced by AI will be ruthlessly eliminated. This unprecedented wave of unemployment is coming at a speed so swift that even governments are unable to formulate any countermeasures. The vast majority of workers, for whom work is the core of the meaning of their lives, will find life meaningless under the weight of unemployment. The resulting social problems will have a profound impact on human society. He predicts that AI and robotics systems with human-like capabilities in various fields will take place by 2027. After three years of devastating decline, the labor market is likely to collapse completely. A 10% unemployment rate is alarming enough, but he’s talking about 99%!
Although not quite so frightening, Jensen Huang, CEO of NVIDIA and a prominent American high-tech figure , also foresees the impact of AI and robotics. In a speech at the Stanford Institute for Economic Policy Research (SIEPR) Economic Summit in March 2024, he predicted that within five years, artificial intelligence will pass all human tests (e.g., bar exams and medical licenses), which would significantly transform education, careers, and knowledge-based work. He also predicted that within ten years, AI computing power would increase a million-fold, giving AI an even greater advantage. At SIGGRAPH, the annual conference for computer graphics and interactive techniques in Denver in August 2024, Huang stated that AI would change everyone’s job, including his own. However, he emphasized that AI significantly enhances human capabilities, not replaces them. Regarding unemployment, Huang mentioned in a July 2025 CNN interview that AI-driven productivity could lead to job losses if humans lack new ideas. However, historically, technology has not only increased employment but also output. He called artificial intelligence the “greatest technology equalizer,” saying it allows non-technical people to easily acquire skills they otherwise wouldn’t have, enabling an “abundance of ideas” for a better future. Meta’s Yann LeCun is equally optimistic, believing that artificial intelligence will transform jobs rather than completely replace them.
Besides this, Adam Dorr of RethinkX predicts mass unemployment by 2045. Geoffrey Hinton, known as the “godfather of artificial intelligence,” believes that “mundane intellectual work” such as call center and paralegal jobs will be replaced, but believes that manual labor such as plumbing is relatively safe. Anthropic CEO Dario Amodei suggests that half of entry-level office jobs could disappear within five years, while OpenAI’s Sam Altman believes that society will adapt and create new roles.
I’ve worked in the US tech industry for 40-50 years, and personally commissioned Grok recently to write JavaScript software. My experience has been that AI is truly amazing. Each time, it gives me nearly a hundred lines of code, each packed with dense logic. When I executed them, the results were incredibly accurate, jaw-dropping! Over the past year, I’ve used seven or eight different AI systems and frequently observed how others were using them. I also feel that AI is bringing incredible changes to various areas of human civilization. However, I still think that the prediction that AI will replace 99% of the world’s jobs within five years is too alarmist. A prediction of 30% within 10 years is more reasonable. Even so, it’s hard to imagine what surviving jobs will look like in 10 years. And with such high unemployment rates, will human society be completely transformed? (2025-09-04 New Jersey)