While the world is heatedly discussing how AI will revolutionize classrooms and achieve “personalized learning,” renowned Finnish education expert Professor Pasi Sahlberg offers a powerful reminder to the sector: “If we think that just by embedding AI to school systems that we have now, then things will get significantly better, I don’t think that’s going to happen”
We have all heard the doubts surrounding traditional education—uniform teaching materials, widening learning gaps, inefficient assessment systems, declining student passion, and a global teacher shortage. But where do the roots of these problems lie? If we are to reform, where should we start? What kind of talent do we need for the future, and how can these qualities be cultivated? At EDUtech Asia 2025 last November, Professor Pasi Sahlberg was undoubtedly the most anticipated keynote speaker. In his speech, he provided a deep analysis of the origins of the modern education dilemma and why the situation continues to worsen. He offered a starting point for policymakers and schools, highlighting the critical factors to watch during transformation. The Unwire team witnessed his compelling presentation and had the privilege of speaking with him afterward to explore his core arguments in depth.
The Root of the Problem: The Formation of a Dilemma
Professor Sahlberg believes the root of global education issues lies in being “outdated.” Today’s massive dilemma was formed by increasing resources to implement “reforms” in the wrong direction.
“Our current education systems were originally designed for an entirely different era and purpose. Previously, society needed a collective workforce with basic literacy; today, society needs future talent that is flexible, creative, collaborative, and passionate about finding their own path. Keeping the old system largely intact is where the problem starts.”
Sahlberg noted that the education sector hasn’t ignored the issues. In fact, the resources invested globally over the past 25 years to “reform” education and improve efficiency have been staggering. The reason for the lack of results, he argues, is that we are trying to solve new-age problems within an old framework. He brilliantly dubbed this quarter-century movement the Global Education Reform Movement, or GERM. Sahlberg believes that this reform movement has not only failed to keep education up to date but has further eroded its foundations. He analyzed four areas where GERM “reforms” have focused, each leading to unintended consequences:
School Competition:
Treating schools as an “industry” and introducing market competition. Originally intended to improve quality through competition, it resulted in skewed resources and inter-school hostility.
Narrow Curricula:
Over-emphasizing literacy and numeracy as the core paths to intelligence. Arts, music, and physical education—vital for holistic development—have been marginalized.
Text-based Accountability:
Using large-scale standardized testing that prioritizes scores and rankings. Students learn under “accountability” systems, bearing heavy pressure. “Studying only for grades” has become a global phenomenon.
De-valuation of Teachers:
Helping students achieve stellar exam results has become the primary KPI for teachers. Teaching has become mechanical, devaluing the profession and discouraging young people from entering the field, worsening the global teacher shortage.
Professor Sahlberg analyzed the “inconvenient data” behind various global education reforms on the EDUtech Asia keynote stage.
Data shows that during the reign of GERM, despite unprecedented investment in funding and technology, average student performance has stagnated, creativity has declined, and the sense of belonging at school is at an all-time low. Sahlberg joked: “If education were an industry, it’s likely the only one that could see zero return on performance for 25 years and still have its budget approved by the board to continue ‘reforming’ in the same failed direction.”
Breakthrough Thinking: “Equality” vs. “Equity”
While Sahlberg acknowledges the importance of technology and funding, he emphasizes that reform must precede investment for resources to be truly meaningful. The starting point for this reform lies in distinguishing between two often-confused concepts: Equality and Equity.
Sahlberg argues that many policymakers believe they are practicing equity when they are actually merely providing “equality of opportunity,” a strategy that frequently backfires. “Giving every student—regardless of personal talent or family wealth—the right to choose an expensive, advanced AI course follows the principle of ‘Equality.’ But is it equitable or inclusive? 25 years of data says no. In fact, it leads to children from disadvantaged backgrounds being more thoroughly marginalized.” He believes that pursuing “Equality” in isolation exacerbates existing imbalances. This creates a polarized system of “elite” schools that parents fight to enter and “weak” schools that no one wants to attend. Students in the latter are not only negatively impacted in their self-confidence by being labeled “mediocre,” but also continue to experience systemic unfairness in resource access throughout their learning journey.
Sahlberg shared the model used in Finland and other Nordic countries, which shares the goal of inclusion but bases resource allocation on Equity. “Simply put, it is about using education to transform an unfair society into a fair one (Equity). In Finland, we call this Positive Discrimination. The goal is to make every school a ‘good school.'”
Finnish education proactively invests more resources into children with poorer learning conditions or lower socioeconomic status to help them rise to a level playing field. Sahlberg believes that only by creating an equal starting point through intentional resource reallocation can social class gaps be eliminated. When education achieves true Equity, inclusion happens naturally, allowing society to foster truly diverse talent.
The True Key: Timing of Investment
Another point Sahlberg emphasized is when resources are invested. “We have known for a long time that the earlier you invest in a child’s education, the more significant the impact and the higher the return.”
Research by Nobel Laureate in Economics James Heckman confirms: the earlier the investment in educational resources, the higher the impact.
He cited Nobel laureate James Heckman’s research from the 1990s, which proved that the ROI for early childhood education is far higher than late-stage remediation. “What is most baffling is that even though this knowledge has been public for 30 years, many systems still don’t prioritize early education. I believe this is the main reason education costs keep rising while effectiveness remains low.” Currently, Finland’s investment in early childhood education is triple that of Australia.
Research by Nobel Laureate in Economics James Heckman confirms: the earlier the investment in educational resources, the higher the impact.
Reform Before Technology
Regarding the hype that “AI will solve century-old education pain points,” Sahlberg does not join the frenzy. He notes that we have experienced this tech-optimism at least twice before.
He recalls: “In the 80s, when ‘one computer per person’ became possible, everyone expected ‘personalized learning’ to arrive. It didn’t. In the 90s, with the rise of the internet, people said global resource sharing would revolutionize education. It didn’t.” He reiterates that AI’s true value will only appear after structural reform. “If we think plugging AI into the old system will change things, I don’t think it will happen.”
However, he sees potential for AI to reduce administrative burdens for teachers. If AI can free up time for teachers to explore diverse assessments and engage in constructive dialogue, that would be its true contribution.
Teacher Training as “Astronaut Training”
Sahlberg offered a radical thought for leaders: teacher training should mirror the design principles of an “Astronaut Training Program”—where 90% of the curriculum is dedicated to preparing trainees for the unexpected.
“In a real classroom, a teacher’s energy and capacity should ideally only require 10% for routine tasks and practical operations,” Sahlberg explained. “The remaining 90% is for managing the interpersonal dynamics with students and other complex, unpredictable situations. This requires adaptive leadership.” He argues that current teacher training systems, as well as the teams leading these educational bodies, are too focused on that 10% of basic tasks. This resistance to change is why the system falls into a state of panic when transformation becomes unavoidable. Perhaps this explains why, with the emergence of AI, the academic community’s most heated discussions still revolve around “teachers’ fear of being replaced,” rather than “how teachers can be empowered.”
Beyond the role of teachers, Sahlberg reminded leaders that a narrow focus on academic performance is outdated. A student’s well-being, resilience, level of engagement, interpersonal relationships, life balance, and openness to knowledge are all critical areas for assessment. “In my home country of Finland, we emphasize the ‘Whole Child Approach.’ We do not view students as mere containers for academic knowledge. Compared to traditional subjects like language and mathematics, we place higher value on the arts, music, physical education, and civic education, focusing on nurturing a child’s sense of happiness and civic values.” He added that his home country has been showing a testament to their resolve, as they stay committed to holistic education as the fundamental premise of learning even when Finland’s position on the PISA (Programme for International Student Assessment) rankings fluctuates.
Trust Our Children
“Please trust our children,” was Sahlberg’s sincere appeal to the educational leaders in the room. “We too often assume young people can’t do this or that, or will go astray. We then conclude: ‘Let me teach you; let me tell you what you need to know.’ The truth is, we severely underestimate them.”
Sahlberg noted that children possess talents and potential that far exceed the level of trust we typically grant them. Their innate hunger for learning and their aspirations for self-growth are qualities that our traditional education system cannot produce. He pointed out that Finland and other leading education systems are currently committed to a profound shift in mindset: moving from viewing students as “objects to be taught” to seeing them as the leads of their own learning (Student Agency). This involves building an educational culture that truly believes in a young person’s ability to participate in—and take charge of—their own learning journey and future development.
Pasi Sahlberg is an internationally renowned Finnish educator, scholar, and author. He formerly served as the Director General of the Center for International Mobility and Cooperation (CIMO) at the Finnish Ministry of Education and Culture. His book, Finnish Lessons: What Can the World Learn from Educational Change in Finland?, was awarded the 2013 Grawemeyer Award.
Editor’s Note
Professor Sahlberg’s insights touch on the deepest pain points of modern education. While proud of the Finnish system, he admits candidly in his book, Finnish Lessons 3.0, that no system is perfect. Our hope is that this interview sparks a meaningful dialogue rather than a debate over who is right.
It is an invitation for us to set aside the biases and defenses tied to our specific roles and positions, so that we may collectively examine the increasingly stark reality of our society. We must ask ourselves: Has mainstream education shifted children’s dreams from “having a happy family,” “contributing to the community,” or “protecting the planet” to simply “getting good grades”?
If so, is that truly what we want to leave for the next generation?
除了位置資訊,新合資企業亦擴大 TikTok 內置 AI 功能相關數據收集權限,當中包括用戶輸入提示字句(prompts)、提問內容,以及 AI 生成內容在何時、何地被建立與使用等操作紀錄。對於利用 TikTok AI 功能創作短片、相片或文字創作者而言,這代表創作過程本身亦會變成平台數據資產,未來如何用於訓練演算法或推薦廣告,現階段仍然缺乏透明度。
在消費性電子市場面臨全球性挑戰的當下,Acer(宏碁)近日在亞太區最大型電競盛事之一 Acer Predator League 2026 印度總決賽前的發佈會上,各區代表交出了令人驚喜的成績單,其中香港市場的表現尤其突出,成為這場亞太發佈會中不容忽視的焦點,當日比賽場地亦設有 Mini CES 展覽館,將拉斯維加斯的 CES 矚目最新電腦及家用產品帶到亞洲觀眾面前。
Microsoft 證實 1 月 Patch Tuesday 更新 KB5074109 令部分 Windows 11 電腦完全無法啟動,顯示「UNMOUNTABLE_BOOT_VOLUME」藍屏錯誤訊息。受影響裝置在黑色畫面上顯示「裝置發生問題,需要重新啟動」,但實際上無法完成啟動程序,必須手動進入 Windows Recovery Environment(WinRE)移除更新才可恢復正常。
影響範圍擴大至實體裝置
Microsoft 表示這次啟動失敗問題主要影響 Windows 11 version 24H2 及 25H2 實體裝置,目前未有虛擬機器受影響報告。雖然公司強調收到報告數量有限,但由於問題令電腦完全無法啟動,對受影響用戶造成嚴重困擾。Microsoft 目前正在研究修復方案及暫時解決方法,但尚未推出針對此問題緊急更新。
除了官方確認問題外,有媒體亦收到用戶回報其他未修復錯誤,包括開機時出現數秒至數分鐘黑屏、桌面背景自動重設為黑色、File Explorer 內 desktop.ini 自訂功能失效等。部分用戶在安裝更新後發現睡眠模式(S3)、Citrix Director 及遠端桌面連線出現異常。這次 1 月更新原本包含 114 個安全漏洞修復,當中包括一個已被主動利用 Desktop Window Manager 零日漏洞(CVE-2026-20805),但多重問題令這次更新成為近年最混亂 Patch Tuesday 之一。
手動移除更新恢復開機
Microsoft 表示受影響用戶需手動進入 Windows Recovery Environment(WinRE)移除 KB5074109 更新。用戶可透過在開機時看到 Windows 標誌後立即按電源鍵強制關機,重複 2 至 3 次後系統會自動進入修復環境。進入 WinRE 後依次選擇「Troubleshoot」>「Advanced options」>「Uninstall Updates」>「Uninstall latest quality update」,輸入管理員帳戶密碼後點擊「Uninstall quality update」按鈕即可移除更新。如無法透過此方法進入修復環境,用戶需要製作 Windows 11 安裝 USB 手指,透過開機選單或 UEFI 設定從 USB 啟動,選擇「Repair my PC」進入修復環境後執行相同步驟。完成移除後建議立即進入「Settings」>「Windows Update」暫停更新,避免系統自動重新安裝這個有問題的更新。
最近在 Social Media 總會看到類似標題:「Google DeepMind 行政總裁:別再實習,學 AI 就夠」、「頂尖 AI 專家告訴你傳統教育已死」。點進去一看,發現內容均引述 Demis Hassabis 在 WEF 的發言。說實話初次看到標題時我也曾心動。誰不想尋找捷徑?但仔細查證後,發現事實並非如此。
斷章取義的狂歡
Hassabis 在名為「The Day After AGI」對談中,確實建議大學生考慮將傳統實習時間用於加強學習 AI 工具。此話前提是 AI 正取代初級重複工作,令傳統實習價值下降。他指出的是職場生態變化下的策略調整,而非要求學生放棄學習基礎知識。
就在論壇舉行前夕,布魯金斯學會發布橫跨 50 國研究報告。報告結論直接指出:以目前狀況觀之,AI 在教育中風險大於益處。研究團隊發現 AI 易用性與「付出少、回報高」的即時反饋,正驅動學生將思考外包給機器,形成認知卸載依賴循環。研究更用了一個刺眼比喻:這種認知退化原本只與老年人大腦衰退有關。
報告共同作者 Rebecca Winthrop 警告,當孩子使用生成式 AI 直接獲取答案,他們便沒有獨立思考,亦沒有學習分辨真假、理解論證或接納不同觀點,因為他們根本沒有與材料深度互動。研究指出缺乏基礎知識的年輕學習者,特別容易接受 AI 生成的錯誤資訊,形成「越不懂越依賴、越依賴越不懂」的危險循環。
放大器而非替代品
我一直認為理解 AI 的最好比喻是「放大器」。它放大有能力者的生產力:你懂數學,AI 幫你算得更快;你懂寫作,AI 幫你產出初稿。但它同樣放大無知者的錯誤:你不懂數學,便無法發現 AI 計算出錯;你沒有邏輯訓練,便無法識別 AI 編造的似是而非論述。
Hassabis 曾言,知道 AI 何時錯誤與知道它何時正確同樣重要。要做到這點,你需要基礎知識、批判性思維與深度學習能力,而這些正是傳統教育核心。布魯金斯報告點出關鍵區別:AI 對大腦成熟的專業人士效果最好,因為他們已具備後設認知與批判性思維,能將 AI 當成認知夥伴。對發育中的年輕人而言,AI 更容易變成認知替代品,反而削弱發展。
那些「精通 AI 就夠」的說法極其危險。它製造虛假安全感,讓人誤以為掌握工具等於掌握能力。工具會更新迭代,亦會被所有人學會。真正持久的競爭力,源於你理解問題本質的深度、質疑答案的能力與從零開始思考的習慣。這些恰恰是傳統教育培養的重點。
結語:聰明地學習而非學習聰明
回到 Hassabis。若要公允概括其教育觀:在 AI 改變工作型態時代,你需要用更聰明方式學習,包括善用工具與保持適應性;但學習深度、批判性思維培養與基礎知識累積,從未像今天這樣重要。他在劍橋說的那句話值得每個焦慮的人記住:真正重要是學會如何學習。AI 的到來不是要替代你思考,而是要測試你有沒有思考能力。
資料來源:
World Economic Forum “The Day After AGI” 對談(2026年1月)
Brookings Institution “A new direction for students in an AI world” 報告(2026年1月)
Apple CEO Tim Cook 的接班時間表持續引發外界揣測,不同報道呈現截然不同的時間預測。《Financial Times》數月前報道指 Apple 正為 Cook 最快於 2026 年初退任做準備,但《Bloomberg》記者 Mark Gurman 今日在其 Power On 電子報中表示,這個時間點「似乎不太可能」。
Apple 硬件工程高級副總裁 John Ternus 被廣泛視為 Cook 的最有力接班人。Gurman 本週稍早報道,Cook 於去年底將 Apple 設計團隊的管理權交予 Ternus,他形容這項人事調動「清楚表明」Ternus 是 CEO 人選的首選。這次職權擴大意義重大,因為 Apple 設計團隊歷來只向公司最高層匯報,包括 Jony Ive(至 2019 年)、Cook 本人(2015 至 2017 年),以及前 COO Jeff Williams(2019 年至去年退休)。
Ternus 自 2001 年加入 Apple,現年 50 歲,自 2013 年起擔任硬件工程高級副總裁。他曾參與 iPhone、iPad、AirPods 等多項產品開發,亦推動 Apple 轉用自家晶片的策略。原本被視為接班熱門的前 COO Jeff Williams 已於 2025 年 7 月退休,令 Ternus 的接班地位更加鞏固。《Bloomberg》早於 2024 年已報道 Ternus 是接班的首選,Apple 當時已加強相關準備工作。
Cook 自 2011 年 8 月出任 Apple CEO,去年已屆 65 歲的一般退休年齡。Gurman 認為 Cook 掌舵 Apple 的時間正逐步邁向尾聲,但他深信接班時間至少仍有數月之遙,並非即將發生。《New York Times》引述消息人士指,Cook 已向高層表達希望減輕工作量的意願,若他退任,預計將出任 Apple 董事會主席。
日本大型電訊公司 KDDI 營業部門員工山本幸一剛傳送企劃書,僅 3 秒後屏幕便彈出「本部長」回覆:「這份提案概念切中要害,但競爭對手案例引用仍停在上季數據,建議參考最新 B 公司案例,我們再來改良。」這位全天候待命、熟記所有最新市場數據的「本部長」,實際上是 KDDI 最新引入的 AI 上司。
AI 上司取代傳統管理模式
KDDI 共部署了 3 位 AI 上司,均以真實本部長(含前任)為原型。系統透過深度學習本尊口頭禪、說話方式與決策邏輯,以似顏繪形式呈現,還原出「三人三樣」獨特領導風格。KDDI 敏捷開發中心副經理河路慶一指出導入核心原因:「高階主管太忙,很難親自過目每份提案,但 AI 能不分時間地點隨時供員工商量。」這套系統自 2024 年秋季開放以來,已覆蓋約 700 名營業同仁,成功讓領導者分身,實現對員工全天候待命支援。KDDI 的 AI 投資不止於此。公司正投資建設亞洲最大 AI 資料庫中心,作為多層次 AI 策略核心基礎建設,涵蓋基礎建設、大型語言模型及 AI 應用程式三個層面。
金融業同步引入 AI 管理
這種 AI 管理模式正逐步在日本企業中推廣。三井住友金融集團於 2025 年夏季導入「AI-CEO」,透過深度學習中島達社長經營會議發言及各類資料,讓員工隨時能諮詢「如果是社長會怎麼想?」,推動經營方針與企業理念深度滲透至組織末梢。
情緒穩定且能給出忠實反饋,是 AI 上司廣受員工好評核心原因。顧問公司 mento 調查顯示,85% 員工表示面對 AI 更容易說出真心話,超過 8 成員工對 AI 反饋持肯定態度,且不排斥接受 AI 指導。更具人性化的是,當 AI 偵測到員工情緒低落或遭遇工作困難徵兆時,會主動向真人主管進言,提供關懷員工具體建議,成為輔助管理者維護團隊關係有力工具。
嚴苛版 AI 教練意外爆紅
並非所有 AI 管理者都走溫和路線。顧問與科技服務公司 Accenture 日本分公司一款員工自研「魔鬼教練」AI 意外爆紅。與常規溫和有禮 AI 不同,這款工具專門模擬「職場最嚴苛現實」,以「咆哮模式」給出尖銳反饋。開發者表示,商場如戰場,既然是模擬體驗,就應還原現實中不允許的嚴苛場景。這種一針見血反饋,反而被員工視為寶貴學習教訓,使用率居高不下。這反向證明只要能助力成長、而非單純情緒化指責,員工其實渴望真實反饋。
重新定義管理者價值
Workday 首席人力資源長辦公室集團總經理 Aashna Kircher 表示,AI 時代企業必須重新調整預期,明確管理者核心責任,培養其關鍵判斷力。在人工智能普及背景下,判斷力、決策力與創造力成為核心訴求,管理者核心價值將轉向「動機賦予」與「職業生涯引導」。日本企業引入 AI 老闆實踐印證了人機協同核心價值:AI 承接重複性管理工作,人類管理者專注於更具溫度與策略價值職責,最終實現 1+1>2 職場效能。全球科技領袖亦積極採用 AI 裝置。Microsoft 行政總裁 Satya Nadella 用 AI 整理通勤時播客內容和會議資料,Apple 行政總裁 Tim Cook 則利用 Apple Intelligence 撰寫冗長電郵,OpenAI 行政總裁 Sam Altman 更將 AI 視為「育兒得力助手」。