運用生成式AI拓展策略思維
一場關於人類洞察與機器智慧協作的真實實驗
Leveraging Generative AI for Expanding Strategic Thinking
A Real-World Experiment in Human Insight & Machine Intelligence Collaboration
聆聽本研究的語音導覽
Listen to the Audio Guide for This Research
開場：尋找策略的優勢
在座的各位策略家，我們都在不斷尋找那份致勝的優勢。因此，在我們最新的研究中，我們決定將當下最熱門的技術之一——生成式AI——放到試驗台上。我們想知道的，不僅是它能否讓流程變得更快，而是它能否真正地讓我們的思維變得更好。而我可以現在就告訴你，結果，和我們預期的不完全一樣。
如果你身處策略領域，你肯定對這種感覺再熟悉不過了，對吧？世界比以往任何時候都更加複雜和不確定。有時，我們傳統的規劃流程感覺就像是拿著上個世紀的地圖，試圖在颶風中航行。這不僅僅是一種感覺，這是全球領導者共同面臨的巨大挑戰。
Opening: The Hunt for a Strategic Edge
Hello everyone. As strategists, we're all constantly on the hunt for an edge. So for our latest research, we decided to put one of the biggest new technologies out there, generative AI, to the test. We didn't just want to know if it could make our processes faster. We wanted to know if it could actually make our thinking better. And I'll tell you right now, the results, they weren't exactly what we expected.
If you're in strategy, you know this feeling deep in your bones, right? The world is just more complex and uncertain than ever. And sometimes our traditional planning processes can feel like we're trying to navigate a hurricane with a map from last century. And look, this isn't just a feeling. It's a massive, well-documented challenge for leaders everywhere.
核心矛盾：強大工具為何束之高閣？
The Paradox: Why Is a Great Tool Left on the Shelf?
幾十年來，情境規劃 一種策略規劃方法，組織透過建構多個關於未來的可能情境故事，來探索和應對不確定性。其目的不是預測未來，而是為了更好地準備應對各種可能性。 一直是我們應對這類不確定性的黃金標準。它是一個絕佳的工具，能打破我們的偏見，建立敏捷性。但矛盾就在這裡：如果它這麼棒，為什麼常常被束之高閣？
答案很簡單：它太慢、太貴，而且耗費的資源是大多數團隊根本無法負擔的。這就是我們想看看AI是否能最終破解的核心問題。
For decades, scenario planning A strategic planning method that organizations use to make flexible long-term plans. It is not about predicting the future, but rather about better preparing for various eventualities. has been our gold standard for dealing with this kind of uncertainty. It's a fantastic tool. It breaks down our biases. It builds agility. So, here's the paradox. If it's so great, why is it so often left on the shelf?
Well, the answer is pretty simple. It's slow, expensive, and it eats up resources that most teams just don't have. And that was the core problem we wanted to see if AI could, you know, finally crack.
實驗設計：給策略家一位AI副駕
核心構想與設置
這讓我們開始思考：如果解決方案不是取代策略家，而是給他們一位智慧副駕駛呢？這就是我們研究背後的核心構想。我們設計了一個實驗，採用「精實情境分析」框架，並接入了一個客製化的AI工具，目標是用真實數據來衡量這種新的人機協作關係。
我們邀請了來自台灣三家成熟企業的32位中高階主管參與4小時密集工作坊，並分離出唯一變數：工具。
- 控制組： 完全回歸傳統。白板、馬克筆，還有一大堆便利貼。
- 實驗組： 同樣的工作坊，但在我們客製化AI工具——由Gemini驅動的「情境分析大師」的協助下進行。
The Experiment: An AI Co-Pilot for the Strategist
The Big Idea & Setup
What if the solution wasn't replacing the strategist, but giving them an intelligent co-pilot? We designed an experiment using a "lean scenario analysis" framework and a custom AI tool to measure if this new human-AI partnership could solve the paradox.
We brought in 32 mid-to-senior level managers for an intensive 4-hour workshop, isolating the single variable of the tools.
- Control Group: Went completely old school. Whiteboards, markers, and sticky notes.
- Treatment Group: Used our custom AI tool, the "Scenario Facilitator Gem," powered by Gemini.
研究引擎：4小時精實情境分析框架
此框架並非要取代深度分析，而是作為一個催化劑，在有限時間內快速啟動高品質的策略對話。它整合了多種策略理論的精髓，作為在不同階段激發特定思維模式的「透鏡」。
聚焦 & 第一原理
"先解構，再綜合"
定義一個清晰的策略問題，並回歸其最根本的驅動因素與假設，避免表面層次的討論。
擴展 & 定義軸線
"擁抱不確定性，設定座標"
腦力激盪出廣泛的外部驅動力量，並選出兩個最關鍵且最不確定的因素，作為建構情境矩陣的軸線。
簡化情境 & 測試策略
"創造故事，測試策略"
為2-3個最具差異性的情境創造簡潔有力的敘事，並系統性地在每個情境中壓力測試組織的現行策略。
行動 & 回饋
"展望未來，回到當下，小步快跑"
綜合所有情境的洞察，識別出「無悔行動」，並將其轉化為具體、可衡量的短期行動計畫。
The Research Engine: The 4-Hour Lean Scenario Analysis Framework
This framework is not a replacement for deep analysis, but a catalyst to jumpstart high-quality strategic conversations under time constraints. It integrates the essence of multiple strategic theories as "lenses" to provoke specific modes of thinking at different stages.
Focus & First Principles
"First deconstruct, then synthesize"
Define a clear strategic question and break it down to its fundamental drivers and assumptions, moving beyond surface-level discussions.
Expand & Define Axes
"Embrace uncertainty, set the axes"
Brainstorm a wide range of external driving forces and select the two most critical and uncertain factors to form the scenario axes.
Simplify Scenarios & Test Strategy
"Create stories, test strategy"
Create brief, compelling narratives for 2-3 of the most divergent scenarios and systematically stress-test the organization's current strategy within each.
Act & Feedback
"Look to the future, return to the present, take small actions"
Synthesize insights from across the scenarios to identify "no-regret moves" and translate them into concrete, measurable, short-term action plans.
研究發現：三個出乎意料的維度
1. 決策風格的轉變：更平衡的策略思維
這可能是我們最大的驚喜。傳統組對 理性 理性決策風格 (Rational Decision-Making Style) 指的是決策者傾向於依賴邏輯、數據和系統性分析來做出選擇。 分析的依賴度驟降，幾乎轉為純粹直覺。而AI組的理性分數只微幅下降。結論是巨大的：AI工具在激發創造性思維的同時，並未迫使他們拋棄分析能力。它幫助創造了一種更平衡、更全面的策略方法。
理性決策風格變化 (平均差異) (?) 此圖顯示工作坊前後「理性」決策風格分數的變化。負值代表對此風格的依賴度下降。傳統組大幅下降(-0.93)，而AI組僅微幅下降(-0.28)，顯示AI在促進直覺思考的同時，也保留了分析的基礎，達成更平衡的思維模式。
詳細數據：決策風格前後測 T檢定結果
| 決策風格 | 組別 | 前測平均 (標準差) | 後測平均 (標準差) | 平均差異 | p-value |
|---|---|---|---|---|---|
| 理性 | 控制組 | 3.82 (0.78) | 2.89 (0.81) | -0.93 | <.001 |
| 理性 | AI組 | 3.79 (0.81) | 3.51 (0.75) | -0.28 | .043 |
| 直覺 | 控制組 | 2.88 (0.85) | 3.95 (0.62) | +1.07 | <.001 |
| 直覺 | AI組 | 2.91 (0.89) | 4.18 (0.59) | +1.27 | <.001 |
2. 效率與產出：更快且更好
AI副駕在效率上完勝，讓流程更實用。更驚人的是，AI組產出的獨特策略選項比傳統組多出 78%。AI不僅讓他們更快，更從根本上打開了對話的大門。
流程效率與滿意度 (ANOVA)
| 項目 | 控制組平均 | AI組平均 | p-value |
|---|---|---|---|
| 時間效率 | 3.40 | 4.41 | <.001 |
| 結構清晰度 | 3.60 | 4.64 | <.001 |
3. 策略產出數量 (ANOVA)
AI顯著增加了探索的廣度（驅動因子）與行動的深度（策略選項）。
| 產出類別 | 控制組數量 | AI組數量 |
|---|---|---|
| 獨特外部驅動因子 | 14.00 | 25.50 |
| 獨特策略選項 | 7.00 | 12.50 |
The Results: Three Surprising Dimensions
1. A Shift in Thinking Style: More Balanced Strategy
This was our biggest surprise. The traditional group's reliance on Rational Rational Decision-Making Style refers to a preference for making choices based on logical, data-driven, and systematic analysis. thinking plummeted, shifting to almost pure intuition. The AI group's rationality score only dipped slightly. The takeaway is huge: The AI tool sparks creative thinking without forcing them to abandon their analytical side, creating a more balanced approach to strategy.
Change in Rational Decision Style (Mean Diff.) (?) This chart shows the change in "Rational" style scores before and after the workshop. A negative value means reliance on this style decreased. The large drop for the Control group (-0.93) vs. the small drop for the AI group (-0.28) is significant. It suggests the AI helps preserve analytical rigor while still encouraging intuitive thinking, leading to a more balanced cognitive outcome.
Detailed Data: GDMS Pre/Post T-Test Results
| Decision Style | Group | Pre-Test Mean (SD) | Post-Test Mean (SD) | Mean Diff. | p-value |
|---|---|---|---|---|---|
| Rational | Control | 3.82 (0.78) | 2.89 (0.81) | -0.93 | <.001 |
| Rational | AI-Augmented | 3.79 (0.81) | 3.51 (0.75) | -0.28 | .043 |
| Intuitive | Control | 2.88 (0.85) | 3.95 (0.62) | +1.07 | <.001 |
| Intuitive | AI-Augmented | 2.91 (0.89) | 4.18 (0.59) | +1.27 | <.001 |
2. Efficiency & Output: Faster and Better
The AI co-pilot was a knockout win on efficiency, making the process feel more practical. More staggeringly, the AI group generated 78% more unique strategic options. The AI didn't just make them faster; it fundamentally blew open the doors of the conversation.
Process Efficiency & Satisfaction (ANOVA)
| Item | Control Mean | AI-Augmented Mean | p-value |
|---|---|---|---|
| Time Efficiency | 3.40 | 4.41 | <.001 |
| Well-Structured | 3.60 | 4.64 | <.001 |
3. Quantity of Strategic Outputs (ANOVA)
AI significantly increased the breadth of exploration (driving forces) and the depth of action (strategic options).
| Output Category | Control Count | AI-Augmented Count |
|---|---|---|
| Unique External Driving Forces | 14.00 | 25.50 |
| Unique Strategic Options | 7.00 | 12.50 |
最精妙的發現：AI的價值並非一體適用
我們最精妙的發現是，AI的價值並非普遍適用，它取決於組織的「準備度」。我們發現組織雙元性 組織雙元性 (Organizational Ambidexterity) 指的是企業能同時高效地「利用 (exploit)」現有業務能力以求穩定，並積極「探索 (explore)」新機會以求創新的雙重能力。 對AI成效有巨大且顯著的提升。但公司一般的創新能力，卻沒有顯著影響。
我們的解讀是：AI是一個流程增強工具，這與組織雙元性——一種管理雙重流程的大師級能力——完美契合。這是工具與企業能力之間的完美握手，從而釋放了真正的價值。
價值釋放模型
The Nuance: AI's Value Isn't Universal
Our most sophisticated finding is that the AI's value isn't universal; it depends on the organization's "readiness." We found that organizational ambidexterity Organizational Ambidexterity is the ability of a firm to efficiently manage its current business (exploit) and simultaneously adapt to new opportunities (explore). gave a huge, significant boost to the AI's impact. But general innovation capability had no significant effect.
Our interpretation: The AI tool is a process enhancement tool. This aligns perfectly with ambidexterity—a master capability in managing dual processes. It's the perfect handshake between the tool and the corporate capability that unlocks the real value.
Value Unlocking Model
總結：人機協作的新框架
總結一下，重要的啟示是什麼？第一，是的，AI可以讓情境規劃變得更快，並產生更廣泛的策略選項。第二，它鼓勵一種更平衡的思維方式，融合了創造性直覺與敏銳的分析嚴謹性。第三，也是應用上最關鍵的一點，最大的效益將在那些已經具備雙元性能力的組織中被釋放。最終，這項研究為我們提供了一個有堅實數據支持的框架，揭示了強大的人機協作策略的真實樣貌。
我想留給你的問題
最終，這項研究不僅僅是為了讓策略變得更快。真正的目標，更大的獎賞，是在我們強大的人類直覺與機器的分析能力之間，鍛造一種更平衡的認知夥伴關係。所以，我想留給同為策略家的你的真正問題，不是你是否應該關注AI，而是：
你的組織，真的準備好釋放它的真正潛力了嗎？
Conclusion: A New Framework for Human-AI Collaboration
Okay, so let's tie this all together. What are the big takeaways here? First, yes, AI can make scenario planning a whole lot faster and generate a much broader set of strategic options. Second, it encourages a more balanced way of thinking, blending that creative intuition with sharp analytical rigor. Third, and this is so critical for actually applying this, the biggest benefits are going to be unlocked in organizations that are already ambidextrous. And finally, what this research gives us is a solid, data-backed framework for what a powerful human-AI collaboration strategy can really look like.
The Question I Want to Leave You With
In the end, this research is about more than just making strategy faster. The real goal, the bigger prize, is to forge a more balanced cognitive partnership between our powerful human intuition and the analytical horsepower of the machine. So the real question I want to leave you with as a fellow strategist isn't if you should be looking at AI. It's...
Is your organization actually ready to unlock what it can really do?