52 Club · 11. 不是另一個讀書會

2049:未來10000天的可能用九組外部證據,重新檢視未來判斷

這一篇把書中的未來想像放回研究與現實資料中檢查:哪些方向已有支持、哪些還缺少條件,以及哪些判斷必須隨新證據更新。

Guided Reading

一次閱讀一個觀點

先看總圖,再依照下方編號逐頁閱讀。每次只顯示目前選取的圖片與解說。

主題總圖

九組外部證據,讓 2049 的想像接受現實檢查

主題總圖

未來想像要能幫助決策,不能只看敘事是否動人。這一篇從預測、生產力、AI 能力、空間運算、隱私、平台治理、生成內容、教育醫療與基因資料九個方向,逐一檢查證據與限制。

  • 先看證據|目前真正觀察到什麼
  • 再看限制|哪些結論還不能成立
  • 最後看更新|什麼新訊號會改變判斷
2049:未來10000天的可能:用外部證據重新檢視總圖點圖開啟原尺寸
01 / 10

讀完後的三個檢查

九項外部檢查共同指出:多數方向並非全對或全錯,而是取決於流程、配套、使用情境與治理。更可靠的做法,是把判斷寫成有條件、可追蹤、會隨證據更新的版本。

證據目前支持到什麼程度?

條件還缺少哪些配套與適用邊界?

更新什麼結果出現時,我應該改變判斷?

Sources

參考資料

文中的編號可點擊查看;此處保留本篇實際使用的完整來源。

  1. [1]

    凱文・凱利著、吳晨編,《2049:未來10000天的可能》,中信出版集團,2025,ISBN 978-7-5217-7518-1。本文分析之附件版本。

    來源為本文分析附件
  2. [5]

    Mellers et al., “Psychological Strategies for Winning a Geopolitical Forecasting Tournament,” Psychological Science, 2014.

    來源 ↗
  3. [10]

    Brynjolfsson, Rock & Syverson, “The Productivity J-Curve,” American Economic Journal: Macroeconomics, 2021.

    來源 ↗
  4. [15]

    European Union, General Data Protection Regulation, 2016.

    來源 ↗
  5. [16]

    Court of Justice of the European Union, C-394/23 judgment on data minimisation, 2025.

    來源 ↗
  6. [17]

    NIST, “Privacy-Preserving Federated Learning” and PETs Testbed, 2025–2026.

    來源 ↗
  7. [22]

    Brynjolfsson, Li & Raymond, “Generative AI at Work,” Quarterly Journal of Economics, 2025.

    來源 ↗
  8. [23]

    Noy & Zhang, “Experimental Evidence on the Productivity Effects of Generative Artificial Intelligence,” Science, 2023.

    來源 ↗
  9. [25]

    Kestin et al., “AI Tutoring Outperforms In-Class Active Learning,” Scientific Reports, 2025.

    來源 ↗
  10. [26]

    OECD, Digital Education Outlook 2026, 2026.

    來源 ↗
  11. [27]

    UNESCO, Guidance for Generative AI in Education and Research, 2023.

    來源 ↗
  12. [28]

    World Health Organization, Ethics and Governance of Artificial Intelligence for Health: Guidance on Large Multi-Modal Models, 2025.

    來源 ↗
  13. [29]

    U.S. Food and Drug Administration, “Artificial Intelligence-Enabled Device Software Functions: Lifecycle Management,” draft guidance, 2025.

    來源 ↗
  14. [33]

    World Health Organization, Human Genome Editing: A Framework for Governance, 2021.

    來源 ↗
  15. [36]

    European Commission, “How the DMA Is Making Smartphones Better,” 2026.

    來源 ↗
  16. [42]

    NIST, AI RMF Generative Artificial Intelligence Profile, AI 600-1, 2024.

    來源 ↗
  17. [44]

    Börjeson et al., “Scenario Types and Techniques: Towards a User’s Guide,” Futures, 2006.

    來源 ↗
  18. [45]

    Meissner & Wulf, “Cognitive Benefits of Scenario Planning,” Technological Forecasting and Social Change, 2013.

    來源 ↗
  19. [46]

    Mellers et al., “Identifying and Cultivating Superforecasters,” Perspectives on Psychological Science, 2015.

    來源 ↗
  20. [47]

    Dell’Acqua et al., “Navigating the Jagged Technological Frontier,” Organization Science, 2026.

    來源 ↗
  21. [48]

    Liang et al., “Holistic Evaluation of Language Models,” Transactions on Machine Learning Research, 2023.

    來源 ↗
  22. [49]

    Becker et al., “Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity,” preprint, 2025.

    來源 ↗
  23. [50]

    Xiong et al., “Augmented Reality and Virtual Reality Displays,” Light: Science & Applications, 2021.

    來源 ↗
  24. [51]

    Koutromanos & Kazakou, “Augmented Reality Smart Glasses Use and Acceptance,” Computers & Education: X Reality, 2023.

    來源 ↗
  25. [52]

    Wang et al., “Integrating AI-Powered Smart Glasses into Digital Health Management,” npj Digital Medicine, 2025.

    來源 ↗
  26. [53]

    Baruh, Secinti & Cemalcilar, “Online Privacy Concerns and Privacy Management,” Journal of Communication, 2017.

    來源 ↗
  27. [54]

    Nouwens et al., “Dark Patterns after the GDPR,” CHI, 2020.

    來源 ↗
  28. [55]

    Selbst et al., “Fairness and Abstraction in Sociotechnical Systems,” FAccT, 2019.

    來源 ↗
  29. [56]

    UK Competition and Markets Authority, AI Foundation Models: Update Paper, 2024.

    來源 ↗
  30. [57]

    Doshi & Hauser, “Generative AI Enhances Individual Creativity but Reduces the Collective Diversity of Novel Content,” Science Advances, 2024.

    來源 ↗
  31. [58]

    Shumailov et al., “AI Models Collapse When Trained on Recursively Generated Data,” Nature, 2024.

    來源 ↗
  32. [59]

    Bastani et al., “Generative AI without Guardrails Can Harm Learning,” PNAS, 2025.

    來源 ↗
  33. [60]

    Wang et al., “The Global State of Clinical Evidence for Artificial Intelligence in Health,” npj Digital Medicine, 2026.

    來源 ↗
  34. [61]

    Vasey et al., “DECIDE-AI: Reporting Guideline for Early-Stage Clinical Evaluation of Decision Support Systems Driven by AI,” Nature Medicine, 2022.

    來源 ↗
  35. [62]

    Martin et al., “Clinical Use of Current Polygenic Risk Scores May Exacerbate Health Disparities,” Nature Genetics, 2019.

    來源 ↗
  36. [63]

    Smith et al., “The BabySeq Project: A Clinical Trial of Genome Sequencing in a Diverse Cohort of Infants,” American Journal of Human Genetics, 2024.

    來源 ↗
  37. [64]

    Holm et al., “The BabySeq Project: Implementing Genomic Sequencing in Newborns,” BMC Pediatrics, 2018.

    來源 ↗

同系列閱讀

也可以從其他角度繼續閱讀

以下文章彼此獨立,沒有固定閱讀順序;選擇現在最感興趣的角度即可。

延伸閱讀書中的九個未來想像延伸閱讀把想像轉成可行動的選擇

Keep Exploring

讀完這一篇,繼續探索下一個問題。

帶著剛剛形成的觀點,看看下一個主題會打開什麼新的視角。

探索更多 52 Club 主題