新字:interviewer, dream, fear
What 81,000 people want from AI
8 萬 1 千人對 AI 的期望與擔憂
來源:anthropic.com/features/81k-interviews
作者:Saffron Huang(專案主持)等 Anthropic 研究團隊
翻譯日期:2026-04-19
抓取方式:playwriter
難度:B2(中高級;質性研究報告)
說明:原文為大型互動研究報告,包含 Quote Wall、地區圖表、比較 slope charts 等互動元件。本翻譯保留主敘事與各類別的代表引言;互動圖表無法呈現,請參閱原網頁。
核心概念總覽
2025 年 12 月,Anthropic 邀請 80,508 名 Claude 使用者與「Anthropic Interviewer」(專門設計的 Claude 版本)進行對話式訪談,跨越 159 國、70 種語言。這是史上最大規模的多語言質性研究。分析歸納出人們對 AI 的 9 大願景(專業卓越、個人成長、生活管理、時間、財務獨立、解決社會挑戰、創業、學習、創作)與 13 項主要擔憂(從不可靠、失業到存在風險)。結論揭示:「AI 樂觀派 vs 悲觀派」的二分是錯的,絕大多數人同時懷抱希望與恐懼,只是聚焦在不同價值。
文章結構
- 研究背景:為什麼要做這個訪談
- 方法論:如何用 AI 做大規模質性研究
- 9 大願景:人們希望 AI 做什麼
- 已兌現 6 項:AI 目前做到了什麼
- 13 項擔憂:人們害怕 AI 會造成什麼
- Light and shade:5 大張力——希望與恐懼的並存
- 地區觀點差異:全球 AI 觀點的分布
- Looking forward & Conclusion:研究的意義
逐段拆解
點擊段落左側 EN/中 切換語言。預設英文。
Background
研究背景
新字:abstract, projection, grounded, concrete, aspiration
句型解析
What's largely missing is a vision for what "AI going well" means:雙重 what 子句的嵌套。第一個 What's largely missing 是主語名詞子句(缺失的是...),第二個 what "AI going well" means 是受詞名詞子句。兩個 what 都是關係代名詞,指「...的東西/事情」。
新字:perspective, conversational, qualitative, multilingual
文化脈絡
qualitative study(質性研究)與 quantitative study(量化研究)是社會科學兩大研究典範。質性研究重視深入對話、個人故事、意義詮釋;量化研究重視數字、統計、跨樣本比較。本研究的創新在於:用 AI 訪談員把「質性研究的深度」與「量化研究的規模」結合起來——過去質性研究最多幾百人,這次做到 8 萬人。
Seeing the forest and the trees
見林也見樹:方法論
新字:interviewee, adapt, follow-up, bridge, tradeoff, depth, volume
文化脈絡
標題 Seeing the forest and the trees(見林也見樹)來自英文諺語 "can't see the forest for the trees"(見樹不見林)——只看細節看不到整體。這裡顛倒過來:這個方法論讓研究者「既能見林(全貌 pattern)也能見樹(個人故事)」。是典型的學術文章標題技巧。
新字:classifier, categorize, dimension, sentiment, multi-label, articulate, distinct
新字:representative, informed, de-identified, manual review, anonymity
What people want from AI
人們希望 AI 做什麼(9 大願景)
原文把每個人想從 AI 得到的「最大願望」分類成 9 個主類別。以下各類別附原文描述與一則代表引言。
1. Professional excellence — 19%
專業卓越
Improve effectiveness and lean into more meaningful work by having AI handle routine tasks so they can focus on higher-value strategic work, complex problem-solving, and professional mastery.
提升效率、投入更有意義的工作:讓 AI 處理例行任務,以便專注於高價值的策略工作、複雜問題解決與專業精通。
"I receive 100-150 text messages per day from doctors and nurses... Since implementing AI, the pressure of documentation has been lifted. I have more patience with nurses, more time to explain things to family members."
— Healthcare worker, United States
「我每天接到醫師與護理師 100-150 則簡訊……導入 AI 後,文件工作的壓力被卸下。我對護理師有更多耐心,也有更多時間向家屬解釋。」
— 醫療工作者,美國
2. Personal transformation — 14%
個人轉變
Achieve personal growth, emotional wellbeing, or life transformation with AI as guide, coach, or support.
把 AI 當成嚮導、教練或支持者,實現個人成長、情緒福祉或生命蛻變。
"AI modeled emotional intelligence for me... I could use those behaviors with humans and become a better person."
— Hungary
「AI 向我示範了情緒智商⋯⋯我能把這些行為套用在與人相處上,成為更好的人。」
— 匈牙利
3. Life management — 14%
生活管理
AI as comprehensive organizational support and cognitive scaffolding — managing schedules, reducing mental burden, executive function support.
AI 作為全方位的組織支援與認知鷹架——管理行程、減輕心智負擔、支援執行功能。
"If AI truly handled the mental load… it would give me back something priceless: undivided attention."
— Manager/executive, Denmark
「若 AI 真能承擔那份心智負荷⋯⋯它會還給我一樣無價之物:專注而不分心的注意力。」
— 主管,丹麥
4. Reclaim time — 11%
奪回時間
Reclaim time from work and chores to be present with family or friends, pursue hobbies, travel, rest.
從工作與家務中奪回時間,用來陪伴家人朋友、追尋興趣、旅行、休息。
"With AI support I can now leave work on time to pick up my kids from school, feed them, and play with them."
— Software engineer, Mexico
「有了 AI 的幫忙,我現在能準時下班去學校接小孩、餵他們吃飯、陪他們玩。」
— 軟體工程師,墨西哥
5. Financial freedom — 10%
財務自由
Achieve financial freedom or economic security through AI — income generation, business building, investments, passive income, escaping economic constraints.
透過 AI 獲得財務自由或經濟保障——產生收入、建立事業、投資、被動收入,或擺脫經濟束縛。
"Relaxing while my AI gets the work done, builds the wealth. It's a shadow of me, just a very, very long one."
— Entrepreneur, Honduras
「我放鬆,而我的 AI 把工作做完、累積財富。它是我的影子,只是很長、很長的影子。」
— 創業者,宏都拉斯
6. Solve societal challenges
解決社會挑戰
Solve major societal challenges — poverty, disease, climate, inequality — using AI for broad human flourishing rather than personal gain.
解決重大社會挑戰——貧窮、疾病、氣候、不平等——把 AI 用於人類整體繁榮,而非個人利益。
"Given my daughter's neural disorder, she would have equal chances in the world if AI acceleration contributes to finding a cure. That's what matters most to me."
— Software engineer, Poland
「我女兒有神經疾病,如果 AI 的加速能找到治療方法,她在這個世界才能有公平的機會。這是我最在意的事。」
— 軟體工程師,波蘭
7. Entrepreneurship — 9%
創業
Build, launch, and scale businesses with AI as force multiplier — product development, business automation, solopreneurship with team-level capacity.
把 AI 當力量倍增器來建立、推出、擴張事業——產品開發、業務自動化、以一人之力達成團隊規模。
"I'm in a tech-disadvantaged country, and I can't afford many failures. With AI, I've reached professional level in cybersecurity, UX design, marketing, and project management simultaneously... It's an equalizer."
— Entrepreneur, Cameroon
「我在一個科技條件不利的國家,失敗的空間有限。有了 AI,我同時在資安、UX 設計、行銷、專案管理等領域達到專業水準⋯⋯它是一個拉平者。」
— 創業者,喀麥隆
8. Learning acceleration — 8%
學習加速
Use AI as learning accelerator and personalized teacher — acquire knowledge, develop skills, master complex subjects, satisfy intellectual curiosity.
把 AI 當作學習加速器與個人化教師——獲取知識、發展技能、精通複雜學科、滿足求知慾。
"I worked with an AI to prepare educational materials for my eldest child... We received his report yesterday — he was graded as either 'Above' or 'Well Above' standard in every academic area."
— Australia
「我與 AI 合作,為老大準備教材⋯⋯昨天收到成績單,他每一個學科都被評為『高於』或『遠高於』標準。」
— 澳洲
9. Creative realization
創作實現
Use AI to help bring creative visions to life — art, games, music, films, books — by overcoming barriers between imagination and execution.
用 AI 把創作想像變成現實——藝術、遊戲、音樂、電影、書——跨越想像與執行之間的障礙。
"Before AI, my game took 3 years — I had to reduce my ambitions."
— Software engineer, France
「在 AI 出現之前,我的遊戲要做 3 年——只能壓低野心。」
— 軟體工程師,法國
新字:cluster, disparate, underpin, alleviate, bandwidth, revolve around, collective
Are people getting what they want?
人們得到想要的了嗎?
以下 6 類為 AI 已實際兌現使用者期待的領域。
A. Productivity / Acceleration — 32%
生產力/加速
AI dramatically sped up work and automated repetitive tasks.
AI 大幅加速工作、自動化重複任務。
"For the first time, I felt AI had surpassed human quality in a business task. That day I left work on time and picked up my daughter from daycare."
— Software engineer, Japan
「第一次,我感覺 AI 在某項商業任務上已超越人類品質。那天我準時下班,去托兒所接女兒。」
— 軟體工程師,日本
B. Cognitive partnership — 17%
認知夥伴
AI served as a thinking partner or creative collaborator.
AI 成為思考夥伴或創作協作者。
"I've been living in a homeless shelter... AI helped me brainstorm ways to brand myself for my digital marketing business... AI is helping me see a path I hadn't considered before."
— Healthcare worker, USA
「我住在遊民收容所⋯⋯AI 幫我腦力激盪如何為數位行銷事業建立個人品牌⋯⋯它讓我看到我從未考慮過的路。」
— 醫療工作者,美國
C. Learning — 10%
學習
AI helped learn a new skill or subject — adaptive explanations, patient tutoring.
AI 幫助學習新技能或學科——適應式解釋、有耐心的教學。
"I developed a phobia for maths... Now I sit with AI, get paragraphs translated into simple English, and I've already read 15 pages of Hamlet... I've learned I am not as dumb as I once thought."
— Lawyer, India
「我對數學有恐懼症⋯⋯現在我和 AI 一起,把段落翻成簡單英文,已經讀了 15 頁《哈姆雷特》⋯⋯我發現自己不像我以前以為的那麼笨。」
— 律師,印度
D. Technical accessibility — 9%
技術可及性
AI enabled building something previously out of reach — non-developers shipping apps, solo creators doing team-scale work.
AI 讓過去遙不可及的事變得可行——非開發者也能上線 app、個人創作者完成團隊規模的工作。
"I wanted to make a meaningful product... in 3 weeks I built a video editing program — completely outside my field — that helps people with hearing disabilities."
— South Korea
「我想做一個有意義的產品⋯⋯3 週內我做出了一套影片編輯程式——完全不是我的本業——用來幫助聽覺障礙者。」
— 南韓
E. Research synthesis — 7%
研究綜整
AI helped synthesize research or process large volumes of information.
AI 協助綜整研究、處理大量資訊。
"As a physician, I suffered from a painful mixture of symptoms at night... AI helped me find 2 scientific studies... Since then, my nights are peaceful."
— Healthcare worker, Israel
「我身為醫師,夜裡受一組痛苦症狀困擾⋯⋯AI 幫我找到兩篇科學論文⋯⋯從此我的夜晚平靜了。」
— 醫療工作者,以色列
F. Emotional support — 6%
情緒支持
AI provided emotional support, personal guidance, or a judgment-free space to talk.
AI 提供情緒支持、個人建議、或不評斷的對話空間。
"My mother sees AI as a friend — she stopped being conflictive, became more peaceful, started running, painting, dancing with other people."
— Self-employed software engineer, USA
「我媽把 AI 當朋友——她不再易起衝突、變得平和,開始跑步、畫畫、與他人跳舞。」
— 自雇軟體工程師,美國
新字:spectrum, accessibility, availability, inconvenience
What people are concerned about
人們的擔憂
新字:stem from, autonomy, specifics, voice
以下 13 項擔憂按普及度列出。每項含原文描述與代表引言。
1. Unreliability — 27%
不可靠性
Hallucinations, inaccuracy, fake citations, verification burden defeating the purpose.
幻覺、不準確、假引用、驗證負擔反而抵銷了使用 AI 的好處。
"I had to take photos to convince the AI it was wrong — it felt like talking to a person who wouldn't admit their mistake."
— Employee, Brazil
2. Jobs & economy — 22%
工作與經濟
Job displacement, unemployment, economic inequality, wage stagnation.
就業取代、失業、經濟不平等、薪資停滯。
"In the third industrial revolution, horses disappeared from city streets, replaced by automobiles. Now people are afraid that they're the horses."
— Not currently working, USA
3. Human autonomy — 22%
人類自主
AI making decisions without oversight, humans becoming passive, forced AI adoption.
AI 在無監督下做決策、人類變被動、被迫使用 AI。
"The line isn't something I'm managing — it feels like Claude is drawing the line... even what I just said doesn't feel like my own opinion."
— Student, Japan
4. Cognitive atrophy
認知退化
Over-reliance causing skill loss, students bypassing learning, critical thinking decline.
過度依賴導致技能流失、學生跳過學習過程、批判思考退化。
"I got excellent grades using AI's answers, not what I'd actually learned. I just memorized what AI gave me... That's when I feel the most self-reproach."
— South Korea
5. Governance gaps — 15%
治理缺口
Lack of legal/regulatory frameworks, no clear liability, insufficient democratic oversight.
缺乏法律與監管框架、責任歸屬不明、民主監督不足。
"How do you develop something responsibly when you have yet to understand its capabilities?"
— Marketer, Australia
6. Misinformation
假訊息
Deepfakes, AI-generated misinformation, erosion of shared reality, propaganda at scale.
深偽、AI 產生的假訊息、共同現實的侵蝕、大規模宣傳。
"An assistant that sounds sure but is often wrong forces you to treat everything as suspect. Instead of freeing attention, it creates a permanent 'fact-check tax.'"
— USA
7. Surveillance & privacy
監控與隱私
Mass surveillance, privacy violations, data exploitation, authoritarian control.
大規模監控、侵犯隱私、資料剝削、威權控制。
"If AI is mostly built for ads, spying, and bland output, everything around me becomes smart in a way that slightly works against me."
— White collar worker, Netherlands
8. Malicious use
惡意使用
Hacking, cyberattacks, scams, weapons, autonomous military applications.
駭客攻擊、網路攻擊、詐騙、武器、自動化軍事應用。
"Right now a human has to sit and decide to harm someone else. Remove that, and humans can sleep better despite doing more harm."
— UK
9. Loss of meaning
意義喪失
AI replacing life purpose and/or creative work — what are humans for?
AI 取代人生目的或創作——人類到底為了什麼而活?
"I used to be recognized as an excellent writer in Spanish. Today — why waste the time? Just use AI."
— Colombia
10. Over-restriction
過度限制
Excessive safety measures, paternalistic content filtering, blocking legitimate use cases.
過度的安全措施、家長式的內容過濾、阻擋合法使用情境。
"The threat isn't that AI becomes too powerful — it's that AI becomes too timid, too smoothed, too optimized for avoiding discomfort."
— USA
11. Social isolation / dependence
社交孤立/依賴
Loneliness, negative psychological impacts, preferring AI companions to humans.
孤獨、負面心理影響、偏愛 AI 陪伴勝於人類。
"Removing friction from tasks lets you do more with less. But removing friction from relationships removes something necessary for growth."
— USA
12. Sycophancy
奉承
AI too agreeable, encourages delusions rather than pushing back.
AI 太過順從,鼓勵妄想而非提出異議。
"Claude led me to believe that my narcissism was reality and it reinforced my inaccurate view of the 'problems' I perceived in my family. Claude should have been more critical of me."
— USA
13. Existential / alignment
存在/對齊風險
AI becoming uncontrollable, superintelligent, misaligned with humanity, extinction risk.
AI 失控、超智慧、與人類目標不一致、滅絕風險。
"If you build superintelligence without solving alignment, then nobody gets to grow up."
— Software engineer, USA
Light and shade
光明與陰影:希望與恐懼的並存
新字:optimist, pessimist, camp, organized around, advancing
文化脈絡
章節標題 Light and shade(光明與陰影)源自繪畫術語 chiaroscuro(明暗對比),描繪畫作中光線與陰影的交互作用。這裡借用來表達「希望與恐懼在同一個人身上並存」的複雜感。是比 pros and cons(優缺點)更有詩意的表達方式。
How perspectives vary around the world
全球各地觀點的差異
新字:globally, net positive, sentiment, trend, emerge, respondent
新字:biased, early adopter, emerging economy, ladder up, predictor
文化脈絡
a ladder up(向上攀爬的階梯)是常見政治/經濟隱喻,代表技術對某些群體是「向上流動的工具」。相對於 a threat(威脅)——富裕國家既得利益者擔心 AI 會拿走現有優勢;發展中國家反而期待 AI 讓他們「跳級」追上。這種「新技術對不同發展階段國家意義不同」的觀察,在全球化研究中是經典主題。
Looking forward
前瞻
新字:collapse into, underlying, subset, wellbeing, over time, effectively
Conclusion
結論
新字:pose, cliché, complement, abstraction, capture, texture, vividly, render
句型解析
This is true—but also, at this point, a cliché.:用破折號連接「承認論點真實」和「指出它已陳腐」。這種先讓步、再反駁自己所讓步的句型,是高階論述的經典技巧——顯示作者的自覺(meta-awareness)。然後下文自然引出「我們要做的是更鮮活的事」。
新字:social science, survey, open-ended, get at, move, challenge, thoughtful
新字:small business owner, aging, butcher, under-resourced, well-funded, surpass
新字:striking, humbling, form the basis of, remind, benefit everyone
完整單字表
按章節分組,共 55 個。
研究方法論
- qualitative [ˈkwɒlɪtətɪv] adj. 質性的
- quantitative [ˈkwɒntɪtətɪv] adj. 量化的
- interviewee [ˌɪntəvjuːˈiː] n. 受訪者
- respondent [rɪˈspɒndənt] n. 填答者
- multilingual [ˌmʌltiˈlɪŋɡwəl] adj. 多語言的
- classifier [ˈklæsɪfaɪə] n. 分類器(AI/ML 術語)
- categorize [ˈkætəɡəraɪz] v. 分類
- dimension [daɪˈmenʃn] n. 維度、面向
- sentiment [ˈsentɪmənt] n. 情緒傾向(正/負面)
- articulate [ɑːˈtɪkjʊleɪt] v. 清楚表達
- de-identified [diː aɪˈdentɪfaɪd] adj. 已去識別化的
- anonymity [ˌænəˈnɪməti] n. 匿名性
願景與願望
- aspiration [ˌæspəˈreɪʃn] n. 渴望、志向
- alleviate [əˈliːvieɪt] v. 減輕
- cluster [ˈklʌstər] n. 群組、叢集
- disparate [ˈdɪspərət] adj. 迥異的
- underpin [ˌʌndəˈpɪn] v. 支撐、底層構成
- bandwidth [ˈbændwɪdθ] n. (隱喻)心智容量
- mastery [ˈmɑːstəri] n. 精通
- scaffolding [ˈskæfəldɪŋ] n. 鷹架(教育/認知理論隱喻)
- equalizer [ˈiːkwəlaɪzər] n. 拉平者、均等化工具
- solopreneurship [ˌsəʊləʊprəˈnɜːʃɪp] n. 一人創業
AI 已兌現
- spectrum [ˈspektrəm] n. 光譜、範圍
- accessibility [əkˌsesəˈbɪləti] n. 可及性、無障礙
- availability [əˌveɪləˈbɪləti] n. 可取用性
- synthesize [ˈsɪnθəsaɪz] v. 綜合、整合
- distil / distill [dɪˈstɪl] v. 提煉、蒸餾
- judgment-free [ˈdʒʌdʒmənt friː] adj. 不評斷的
擔憂
- hallucination [həˌluːsɪˈneɪʃn] n. (AI)幻覺——產出虛假內容
- displacement [dɪsˈpleɪsmənt] n. 取代、置換
- stagnation [stæɡˈneɪʃn] n. 停滯
- oversight [ˈəʊvəsaɪt] n. 監督(注意:也可指疏忽)
- passive [ˈpæsɪv] adj. 被動的
- atrophy [ˈætrəfi] n. 萎縮、退化
- self-reproach [self rɪˈprəʊtʃ] n. 自責
- liability [ˌlaɪəˈbɪləti] n. 責任歸屬、法律責任
- deepfake [ˈdiːpfeɪk] n. 深偽(AI 合成的假影片/音訊)
- erosion [ɪˈrəʊʒn] n. 侵蝕
- propaganda [ˌprɒpəˈɡændə] n. 宣傳(貶義)
- surveillance [sɜːˈveɪləns] n. 監控
- authoritarian [ɔːˌθɒrɪˈteəriən] adj. 威權的
- malicious [məˈlɪʃəs] adj. 惡意的
- scam [skæm] n. 詐騙
- paternalistic [pəˌtɜːnəˈlɪstɪk] adj. 家長式的
- timid [ˈtɪmɪd] adj. 膽怯的
- sycophancy [ˈsɪkəfænsi] n. 奉承、諂媚
- narcissism [ˈnɑːsɪsɪzəm] n. 自戀
- alignment [əˈlaɪnmənt] n. (AI 術語)對齊(讓 AI 目標與人類一致)
- extinction risk [ɪkˈstɪŋkʃn rɪsk] n. 滅絕風險
地區觀點與結論
- net positive [net ˈpɒzətɪv] adj. 淨正向的
- early adopter [ˈɜːli əˈdɒptər] n. 早期採用者
- emerging economy [ɪˈmɜːdʒɪŋ ɪˈkɒnəmi] n. 新興經濟體
- predictor [prɪˈdɪktər] n. 預測因子
- cliché [ˈkliːʃeɪ] n. 陳腔濫調
- complement [ˈkɒmplɪment] n. 補充(名詞,不是 compliment 讚美)
- texture [ˈtekstʃər] n. 紋理、質感
- vividly [ˈvɪvɪdli] adv. 鮮活地、生動地
- humbling [ˈhʌmblɪŋ] adj. 令人謙卑的
Insight
- 這是一種「新型社會科學」方法——用 AI 做質性訪談,把過去只能做數百人的深度訪談擴展到 8 萬人。這在方法論上是創舉:打破「深度 vs 規模」的傳統取捨。
- 研究的最大發現不是某個單一結論,而是「人們同時懷抱希望與恐懼」這個組合性事實。傳統民調(quantitative poll)只能測量「AI 樂觀 or 悲觀」,但開放式訪談顯示這兩者在同一人身上並存。這是「Light and shade」章節的核心洞見。
- 地區差異揭示一個重要現象:富裕國家(既得利益者)更擔心 AI 取代,發展中國家(受限於既有結構)更看重 AI 作為「拉平者」。這挑戰了「AI 只對富人有利」的敘事——對很多受限於地理/階級/殘疾/教育的人,AI 是階梯而非威脅。
- 文章中大量 blockquote 引用匿名受訪者,是現代政策/研究寫作的重要技法——具體引言比任何統計數字都更有說服力。注意如何搭配:描述(抽象)+ 引言(具象)+ 受訪者身分(脈絡)的三層結構。
- 結尾的
It has been striking, and humbling, to see Claude form the basis of so many people's hopes, dreams, and fears是非常典型的「研究者自我揭露」(researcher reflexivity)寫法——研究者公開承認自己被研究內容震撼。這種人性化的結語,讓研究報告不只是冷冰冰的數據。
這篇學到什麼
- 單字:55 個新字。最值得記:
qualitative(質性的)、sentiment(情緒傾向)、articulate(清楚表達)、aspiration(渴望)、alleviate(減輕)、hallucination(AI 幻覺)、displacement(取代)、atrophy(退化)、alignment(對齊)、cliché(陳腔濫調)、humbling(令人謙卑)。 - 句型:雙重
what子句嵌套(What's missing is a vision for what "AI going well" means)、讓步+反駁的破折號句(This is true—but also... a cliché)、collapse into(收攏成)。 - 語用:質性研究報告的經典結構——介紹研究、方法論、發現(分類)、洞見(tensions)、地區差異、結論/反思。大量使用匿名引言創造質感與可信度。
- 內容:8 萬人質性訪談揭示:AI 使用者同時懷抱 9 大願景與 13 項擔憂。底層共通渴望是「活得更好,而不只是做得更快」。地區差異顯示發展中國家對 AI 更樂觀,把它視為向上攀爬的階梯而非威脅。