G P 報告稱受訪高層 100% 都有使用 AI,但 73% 指至少部分 AI 投資回報令人失望或未達預期 [2][3]。 AI 投資壓力來自幾方面:積極部署降溫、約七成高層或在今年看不到生產力增長時削減投資、69% 指員工花更多時間覆核 AI 產出 [2][3]。

Create a landscape editorial hero image for this Studio Global article: What does G-P’s latest AI at Work Report reveal about falling executive confidence in AI ROI, including the finding that 73% of executives s. Article summary: G-P’s 2026 AI at Work Report shows a clear shift from AI enthusiasm to ROI pressure: executives are still using AI, but many now say results are underwhelming, costly to validate, and harder to connect to real business v. Topic tags: general, general web. Reference image context from search candidates: Reference image 1: visual subject "With 73% of executives reporting at least some of their AI investments fell short of expectations over the past 12 months, this year's report" source context "The AI Reckoning: 73% of Executives Report Underwhelming ROI from AI Efforts as Focus Shifts from Hype to High-Stakes Pr" Reference image 2: visual subject "With 73% of ex
Studio Global AI
Use this topic as a starting point for a fresh source-backed answer, then compare citations before you share it.
G P 報告稱受訪高層 100% 都有使用 AI,但 73% 指至少部分 AI 投資回報令人失望或未達預期 [2][3]。
G P 報告稱受訪高層 100% 都有使用 AI,但 73% 指至少部分 AI 投資回報令人失望或未達預期 [2][3]。 AI 投資壓力來自幾方面:積極部署降溫、約七成高層或在今年看不到生產力增長時削減投資、69% 指員工花更多時間覆核 AI 產出 [2][3]。
報告亦揭示「表演式生產力」風險:88% 高層擔心員工用 AI 扮忙而未創造價值;82% 又稱 AI 降低了他們對人類員工價值的評估 [3]。
繼續“AWS Kiro 更新:AI 寫 Code 前,先用 Requirements Analysis 驗清 Spec”以獲得另一個角度和額外的引用。
開啟相關頁面對照「Panasonic 5,000億日圓押注 AI 基建:數據中心電池點樣救 EV 逆風?」交叉檢查此答案。
開啟相關頁面Investments in artificial intelligence are growing, but some research suggests CEOs may be finding that the new technology is leaving many feeling that it’s falling short. Nearly three in four (73 per cent) executives say their return from investments on ar...
G-P’s 2026 AI at Work Report reveals a global pivot from blind AI adoption to demands for high-stakes accountability and tangible business value REMOTE FIRST COMPANY, May 12, 2026 – G-P (Globalization Partners), recognized as the undisputed leader in global...
achieving AI value at scale—a measure of how tough the full AI transformation is. Fully 60% of companies are not achieving material value at all, reporting minimal revenue and cost gains despite substantial investment. Another 35% (13 percentage points more...
Workday research finds AI boosts productivity, but rework erodes ROI, as errors, trust gaps, and outdated roles prevent organizations from capturing full value. Workday, an enterprise AI platform, recently released new research that found AI is delivering p...
G-P 嘅報告並唔代表企業準備放棄 AI。相反,受訪高層全數都話自己有用 AI [3]。轉變在於,企業已經過咗「先部署再算」嘅階段,開始要求 AI 對業務成果有交代。
呢個氣氛同上一輪明顯唔同。G-P 2025 年《AI at Work Report》仍然強調加速採用:91% 高層表示正積極擴大 AI 計劃,74% 指 AI 對公司成功至關重要 [14]。到 2026 年,AI 仍然普及,但討論焦點已轉向問責、壓力測試同投資回報證明 [
3]。
其他研究亦指向類似嘅「價值落差」。Boston Consulting Group(BCG)曾指出,60% 公司未能喺大規模 AI 應用中取得實質價值;另外 35% 雖然見到部分回報,但承認推進得未夠遠、未夠快 [7]。麥肯錫亦發現,92% 公司計劃未來三年增加 AI 投資,但只有 1% 領袖認為自己公司已成熟到可以將 AI 完全整合入工作流程,並帶來重大業務成果 [
13]。
AI ROI 令人失望,其中一個原因係:工具可能加快咗某一步,但將成本轉移到之後嘅覆核、修正同風險管理。G-P 指,69% 高層表示員工花更多時間監察、審閱或更新 AI 生成內容 [3]。
現實入面,AI 可以好快生成草稿、答案、程式碼或者摘要;但如果之後要由人檢查準確性、重寫語氣、處理錯誤、確認合規,咁節省出嚟嘅時間就未必真係變成淨生產力。
重點係,產出多唔等於效率高。真正要計 ROI,應該計「淨生產力」:AI 慳咗幾多時間,減去人手覆核、改錯、重寫同更新所花嘅時間。Channel Insider 引述 Workday 研究亦有類似觀察:AI 帶來嘅時間節省,可能會被修正錯誤、重寫內容同反覆核對輸出所抵消 [11]。
G-P 報告另一個值得留意嘅位,係管理層開始擔心 AI 使用變成一種表面 KPI。調查指,88% 高層擔心員工用 AI 去顯得忙碌,或者只係配合公司對 AI 使用嘅要求,但未有產生有意義成果 [3]。當中 47% 表示非常或極度擔心呢種情況已經發生 [
3]。
呢點對公司量度 AI 成效好重要。如果只睇工具登入次數、prompt 數量、AI 草稿數目,或者員工自報「我有用 AI」,好容易將活動量誤當成價值。呢啲數字可以證明有人用工具,但唔能夠證明工作更好、更快、更安全,或者更賺錢。
呢個矛盾值得企業正視:AI 唔係令人成為多餘,而係令「人點樣介入」變得更關鍵。好多時候,真正令 AI 有用嘅,正正係人嘅判斷、背景知識、風險意識同質量把關。
麥肯錫嘅職場 AI 研究亦主張,公司應聚焦能夠幫員工處理日常工作的實際應用,並將 AI 同可量度成果連接起來,而唔係將 AI 當成一個孤立嘅科技部署項目 [13]。
G-P 呢份報告最實用嘅啟示係:企業唔應該只量度 adoption,而要量度 outcome。更成熟嘅 AI 計分板,至少應包括:
講白啲,公司要停止只問「員工有冇用 AI」,改為問「AI 有冇改善真正重要、經驗證嘅工作」。
G-P 2026 年《AI at Work Report》唔係一個反 AI 故事,而係一個問責故事。同一份調查一方面顯示 AI 使用已經普及,另一方面亦顯示 73% 高層對至少部分 AI 投資感到失望,而接近七成可能會喺目標未達時收縮開支 [2][
3]。
由於報告建基於高層調查回應,唔應該被解讀成 AI 已被審計證明失敗。不過,佢清楚反映一件事:AI 嘅舉證責任已經轉移。下一階段嘅職場 AI,重點唔再係部署幾多工具,而係能否交出可量度、可信、經人判斷驗證嘅商業價值。
AWS Kiro Requirements Analysis:AI 落手寫 Code 前先驗清需求
Over the next three years, 92 percent of companies plan to increase their AI investments. But while nearly all companies are investing in AI, only 1 percent of leaders call their companies “mature” on the deployment spectrum, meaning that AI is fully integr...
New research from Globalization Partners reveals global leaders use AI for nearly half of their work – and many are willing to switch to AI-friendly companies to gain access to the technology REMOTE FIRST COMPANY, April 8, 2025 – G-P (Globalization Partners...