面對人工智慧(Artificial Intelligence, AI)，南非已故總統尼爾森•曼德拉一句有名的話，可以做為大家的借鑑。
‘May your choices reflect your hopes, not fears.’ (～Nelson Mandela)
願你的選擇反映出你的希望，而不是你的恐𢣷。(~尼爾森•曼德拉)
……Life is about choices, big and small and the decisions we make to take them. We do know that AI will, at some point, dominate every aspect of human life. An Artificial General Intelligence(AGI) will impact every single country, industry and person. The world will not be the same. If we choose to grow with AI, we must decide how to select the best possible future for humanity.
生活是大大小小的選擇。AI對人類的影響將鋪天蓋地。有自我學習能力的通用人工智慧(Artificial General Intelligence, AGI)，將影響每一家企業，每一個國家，每一個人。如果我們選擇和人工智慧一起成長，那麼我們就得為人類的未來慎選最好的將來。
Short-term milestones: 2022-2025
- Stop writing guidelines and start agreeing
- Create principles and standards using broad-ranging multiple-stakeholder group
- Improve policymakers technical knowledge
- Make AI more fair by removing bias from AI data sets
- Start controlling AI by risk management & auditing
短期目標2022-2025：
大家儘快同意一套AI 管理的原則和標準。
制定政策的人要加強自身對AI的認識。
防止AI學習大家普遍存在的偏見。
開始對AI進行風險管理及稽核。
Medium-term milestones (2025-2028)
- Police AI to prevent misuse and protect people
- Create an NPT-style treaty for laws
中程目標2025-2028
檢查AI防止其濫用以至危害人類。
針對AI，簽訂類似控制核武器的「核不擴散條約」(Nuclear Non-Proliferation Treaty, NPT)。防止AI 科技落入邪惡團體的手𥚃。
Medium to long-term milestones 2025-2030
- Educate the world to understand and to work alongside AI
- Rethinking the existing workforce
- Manage economic consequences of Artificial Intelligence
中長程目標 2025-2030
讓世人更了解AI是什麼。
解決AI取代人力所產生的失業及相關轉型的問題。
管理AI所產生的經濟影響。
If AI has a similar catastrophic effect on the job market, we need to be ready for the transformation. Starting by identifying the most at-risk roles and mapping skills(and people) to complementary roles will help direct our investment in reskilling and inward investment. This planning is a simple first step that also helps aligns educational efforts at school and university. The next step is to commit reskilling people just in advance of when the new skills are needed…
AI 會取代某些工種的人力。我們必須提前佈署，改變學校教育的內容。並提早標示出目前在職場上會為受到影響的人，為他們提供學習新技能的教育訓練，以方便轉職。
……experts tend to agree that as AI becomes a mature technology, it will displace physical jobs with virtual labor at a gallop. On the credit side, AI will boost overall productivity, so tax receipts from Corporate Tax will increase. The new jobs created by AI will , on average, be more skilled than those lost, so be more highly paid.
However, there will be consequent redirection in overall tax revenues globally…
專家傾向於同意，當AI成熟，勞務的工作會被迅速取代。生產力提高，公司税增加，AI雇員也是比較高的所得(繳更多的個人所得稅。
科技巨頭傾向於把公司設在公司稅較低的海外，以規避在母國的公司稅。
Problems will start when we see the job loses from structural economic change brought about by AI. In OECD countries, Corporate Tax is, on average, only 3.5% of total tax revenue(rising to 9%-10% in the UK or Germany). Even using the higher 10% figure, revenue from Corporate Tax is dwarfed by 24% of income from Employment Taxes and Sales Tax. So the real challenge for governments will be the domino effect of losing both Corporate Taxes AND the reduction in employment and sales taxes due to mass unemployment…
當AI造成大量失業的時候，政府將面臨財政收入短少的挑戰。在開發中國家，平均公司稅只佔政府稅收3.5%，就算是用英國和德國比較高的10%來看，比起佔政府整體稅收24%的個人所得稅及消費稅，是小巫見大巫。
對政府而言，既收不到增加的公司稅(規避到海外了)，也因為大量的失業以及因此撙節開支的減少消費，會大幅短收個人所得稅和消費稅。
……overall global GDP is likely to increase but the wealth gap between rich(employed, AI-skilled, working in Technology) and poor(unemployed/with no AI skills/working in repeated/process-driven jobs)will widen. Growth in GDP will be more significant in countries where Mega-Tech is based, deepening global inequity, GDP increases do not necessarily lead to increases in government tax collection in all counties.
AI會增加全球的GDP，並不是每個人利益均沾。有AI技術的人比起沒有的人，會變得更富有。
有大科技公司的國家GDP會增加，拉大和其他國家的距離。至於GDP的增長對政府公司稅的貢獻，對每個國家的影響也是不同的(因為避稅)。
As we spend so much time at work, we often build shadow ‘work families.’ We make friends with colleagues and customers, and many romantic relationships start with an introduction at work or through friends at work. Work is vital for many people to derive their sense of purpose. But when your life’s purpose comes from work, when that work goes away, it can be devastating. Research shows that unemployed people have much worse mental health than those in work. Unemployment leads to higher levels of substance abuse, crime, and lower life expectancy. Families with a multi-generational history of unemployment have even worse issues in these areas.
我們花很多時間在工作上，直接或間接讓我們感受到生命的意義。根據研究，失業是非常致命的打擊。失業的人，有更高的藥物濫用、犯罪，而且壽命更短。跨代的連續性失業，狀況更糟(譬如在所謂的「鐵鏽區」)
……But people want to know if AI replaces their jobs — how will they live. The most commonly held approach to funding the AI jobs gap is to introduce Universal Income (UI). Universal Income is where everyone of working age is paid a fixed salary with no obligation to do anything in return…
如果AI取代了工作，那麼失業者的生活何以為繼呢？「全民基本收入」是經常被提起的的概念和做法：每位足齡可以進入職場的人，每個月都可以從政府收到一筆收入來支應日常生活所需，但他不必一定得去工作。
(有各種實驗在不同的國家實驗，方式和結果不太相同，相當有爭議性。芬蘭的實驗結果頗正面。以台灣而言，尚且不談是否因此鼓勵不勞而獲，就光「全民基本收入」的支出，就完全不是現在的財政收入可以負擔的，必定得大幅加稅，以目前的稅制就是加重薪資所得者的負擔。如果那是均富的手段，手上有很多資產的富人又相對付出得少。並不公平。況且健保收入的不足以及人口老化所造成的勞退準備的缺口，那是更迫切的危機。)
……If we don’t need to work, how do we give ourselves purpose? How do we spend the 1,700 hours per year of free time？
……There are some challenges to humanity that look impossible to solve. With tome and space given to us by AI, we could ask better questions of ourselves and look to conquer these challenge.
如果我們不再需要工作，那麼我們怎麼為自找到生命的目的？我們怎麼去利用每年多出來的1700小時呢？
人類面臨一些看起來無法解決的挑戰。AI讓我們騰出時間和空間，更可以追尋生存的目的，並逐步克服這些挑戰和問題。
讀後
- 針對AI的巨大能力，全球必須迅速制定並實施道德的共識。
- 必須管控AI的風險，而這不能留給唯利是圖的科技巨頭去隨意為之。
- AI對失業的潛在影響可能很大，要提前部署預應的措施。
- 對個人、公司或國家而言，AI已是大勢所趨，要有隨之一起蛻變成長的心理準備。
- 工業革命縮短了大家的工時，AI也會。我們會有更多的時間，反問並追求自我存在的意義。
- AI 會拉開富國和貧國之間的差距。人和人之間也是。在AI的時代，要順勢而為，終身學習，才能避免變成數位貧民或者數位難民。
*: Catriona Campbell, AI by design, 2022, CRC press
2025/12/12 AI by design Damakey
