Modern Lens · 科技現代視角 · 科技

AI and the Five ElementsAI 與五行

Understanding Artificial Intelligence Through Ancient Chinese Philosophy透過中國古代哲學理解人工智慧

10 min read · Reviewed September 2026閱讀時間約 10 分鐘 · 2026 年 9 月校閱

Ink-wash artwork of a circuit path meeting a five-node gold ring — Used as a systems model, the five phases are a surprisingly good thinking tool. Plate 01 作為系統模型，五行是一個出奇好用的思考工具。圖版 01

Artificial intelligence is transforming our world at breakneck speed. But as we build increasingly sophisticated AI systems, we face a surprising realization: the questions we are asking about AI — What is consciousness? How do systems maintain balance? When does growth become destructive? — are the same questions that Daoist philosophers explored thousands of years ago using the framework of Yin Yang and the Five Elements (Wu Xing).人工智慧正以驚人的速度改變我們的世界。然而，隨著我們建構出日益複雜的 AI 系統，我們面臨一個令人驚訝的體悟：我們針對 AI 所提出的問題——什麼是意識？系統如何維持平衡？成長何時會演變成破壞？——正是數千年前道家哲學家利用陰陽與五行（Wu Xing）框架所探索的相同課題。

This page maps the five elements — Wood, Fire, Earth, Metal, and Water — onto the core components of modern AI systems. The goal is not to force an artificial fit, but to use an ancient systems-thinking framework to gain fresh perspective on the technology shaping our future.本頁將五行——木、火、土、金、水——對應到現代 AI 系統的核心元件。其目的並非強行牽強附會，而是透過古老的系統思考框架，為這項塑造我們未來的科技帶來全新的觀點。

The Five Elements of AIAI 的五行

Just as the Daoist Five Elements describe five modes of transformation in nature, we can identify five foundational components of any AI system. Each maps to an element not by surface resemblance, but by functional role — what it does within the system.正如道家的五行描述了自然界中的五種轉化模式，我們也能在任何 AI 系統中識別出五個基礎元件。每個元件對應到一個五行，並非出於外表相似，而是依據其功能角色——也就是它在系統中所發揮的作用。

Wood (木) — Data Growth · Expansion木 — 資料生長 · 擴張

In nature, Wood represents growth, vitality, and the branching expansion of life. Trees grow by absorbing nutrients from soil and sunlight, constantly reaching outward.在自然界中，木代表生長、活力以及生命的枝條擴充。樹木透過吸收土壤中的養分與陽光來成長，不斷向外延伸。

In AI, data is Wood. Data is the raw material that feeds the system, the nutrient that enables growth. Just as a tree withers without water and sunlight, an AI model stagnates without quality data. The explosive growth of the internet — billions of text passages, images, videos — is the forest from which modern AI draws its nourishment.在 AI 中，資料即是木。資料是供給系統運作的原料，也是促進成長的養分。正如樹木沒有水和陽光就會枯萎，AI 模型沒有優質資料也會停滯不前。網際網路的爆發性成長——數十億段文字、圖像、影片——正是現代 AI 汲取養分的森林。

Like Wood in the generating cycle (Water generates Wood), data is produced by the flow of human activity (Water) and, in turn, fuels computation (Fire). And just as overgrown forests can become unsustainable, data hoarding — collecting personal data without consent or purpose — is Wood growing out of control, needing the pruning hand of Metal (algorithms and regulation).正如相生循環中的木（水生木），資料是由人類活動（水）的流動所產生，進而滋養運算（火）。正如過度茂密的森林會變得無法永續，資料囤積——在未經同意或無明確目的下收集個人資料——就像是失去了控制的木，需要金（演算法與法規）的剪修之手。

Fire (火) — Computation Transformation · Energy火 — 運算轉化 · 能量

Fire in Wu Xing theory represents transformation, heat, and the active conversion of energy. It is the element of rapid change and intense activity.在五行理論中，火代表轉化、熱能以及能量的動態轉換。它是代表快速變化與劇烈活動的元素。

In AI, computation is Fire. The training of a large language model is an act of pure transformation — billions of parameters adjusted through trillions of calculations, converting raw data (Wood) into learned representations. The GPUs and TPUs running these calculations literally generate heat, consuming enormous electrical energy.在 AI 中，運算即是火。大型語言模型的訓練是一種純粹的轉化行為——透過數兆次計算來調整數百億個參數，將原始資料（木）轉化為學習到的表達特徵。執行這些計算的 GPU 與 TPU 確實會產生熱能，並消耗龐大的電能。

The generating cycle says Wood feeds Fire: more data enables more powerful computation. But Fire also needs control. The controlling cycle says Water controls Fire — meaning memory and storage infrastructure (Water) must scale with computation, or the system becomes unbalanced. This is why AI companies invest simultaneously in compute clusters and data centers: Fire and Water must grow together.相生循環指出木生火：更多資料能實現更強大的運算。但火也需要克制。相剋循環指出水剋火——意味著記憶體與儲存基礎設施（水）必須隨運算規模同步擴充，否則系統就會失衡。這就是為什麼 AI 公司會同時投資於算力集群與資料中心：火與水必須共同成長。

The current industry trend toward ever-larger models — "just add more compute" — is Fire burning without restraint. Daoist wisdom would counsel: when Fire is excessive, the system consumes its own resources. Efficiency, pruning, and smaller purpose-built models are the Earth that tempers Fire's intensity.當前產業趨勢傾向追求越來越大的模型——「只要增加更多算力即可」——這正是火不受節制地燃燒。道家的智慧會建議：當火過旺時，系統會消耗自身的資源。效率、剪枝以及精簡的專用模型，正是調和火之劇烈能量的土。

Earth (土) — Integration Stability · Nourishment土 — 整合穩定 · 滋養

Earth represents the center, the ground that holds all things together. It is the element of stability, mediation, and nourishment — the soil through which all other elements interact.土代表中央，是凝聚萬物的基石。它是象徵穩定、調和與滋養的五行——所有其他五行交會互動的土壤。

In AI, integration is Earth. This includes model architecture, the structural design that determines how data flows through the system; middleware that connects models to applications; and the human-in-the-loop systems that keep AI aligned with human needs. Earth is what prevents a powerful model from becoming a loose cannon.在 AI 中，整合即是土。這包括模型架構（決定資料如何在系統中流動的結構設計）、連接模型與應用程式的中間件，以及讓 AI 保持契合人類需求的「人機協同」（human-in-the-loop）系統。土能防止強大的模型失去控制。

Earth is generated by Fire — meaning computation produces integration. Every training run refines the architecture, every deployment teaches us how to better structure the system. And Earth generates Metal — integration produces the refined algorithms that make the next generation more efficient.火生土——意味著算力產生整合。每一次訓練都提煉了架構，每一次部署都教會我們如何更好地構建系統。而土生金——整合產生了精煉的演算法，讓下一代系統更加高效。

When Earth is weak, AI systems become fragmented: brilliant models that cannot be deployed safely, powerful tools that lack user-friendly interfaces, or systems that work in the lab but fail in production. The current focus on AI safety and alignment is essentially a focus on strengthening Earth.當土弱時，AI 系統就會變得支離破碎：無法安全部署的卓越模型、缺乏友善使用者介面的強大工具，或是僅能在實驗室運作卻無法在生產環境發揮作用的系統。當前對 AI 安全與對齊的關注，本質上就是加強土的力量。

Metal (金) — Algorithms Precision · Structure金 — 演算法精準 · 結構

Metal in Wu Xing represents structure, refinement, and the cutting away of impurities. A sword is Metal — it separates what is essential from what is not.Wu Xing 中的金代表結構、精煉與去蕪存菁。寶劍即是金——它能將精髓與雜質剔除分離。

In AI, algorithms are Metal. The mathematical structures — transformer architectures, attention mechanisms, loss functions, optimization algorithms — are the precise, refined tools that shape raw computation into useful intelligence. Like a blacksmith forging a blade, the algorithm designer hammers raw mathematics into a tool that cuts through complexity.在 AI 中，演算法即是金。數學結構——Transformer 架構、注意力機制、損失函數、最佳化演算法——是將原始算力塑造為有用智慧的精準精煉工具。就像鐵匠鍛造刀刃一樣，演算法設計師將原始數學打造成剖析複雜問題的工具。

Metal controls Wood — algorithms determine how data is used. Good algorithms extract signal from noise; poor algorithms drown in data without learning. And Metal is generated by Earth — solid integration enables algorithmic innovation. The transformer architecture, for instance, emerged from years of integration experience with RNNs and CNNs.金剋木——演算法決定了資料如何被使用。好的演算法能從雜訊中提取訊號；差的演算法則沉溺於資料中而無法學習。且土生金——穩固的整合促進了演算法的創新。例如，Transformer 架構正是源於多年來對 RNN 和 CNN 的整合經驗。

The current push toward efficient AI — smaller models, quantization, pruning, distillation — is Metal doing what it does best: cutting away the unnecessary to reveal the essential. A 7B parameter model that performs as well as a 70B model is Metal at its finest.當前推動高效 AI 的努力——更小的模型、量化、剪枝、蒸餾——正是金發揮其專長：去蕪存菁以顯現本質。一個性能媲美 70B 模型的 7B 參數模型，就是金之極致展現。

Water (水) — Memory Storage · Depth水 — 記憶儲存 · 深度

Water represents depth, storage, and the potential that precedes manifestation. It is the most Yin of the elements — quiet, receptive, holding everything in potential.水代表深度、儲存以及顯化前的潛能。它是五行中最具陰性的存在——安靜、包容，將一切蘊藏於潛能之中。

In AI, memory and storage are Water. This includes the training data repositories, the vector databases powering retrieval-augmented generation (RAG), the weights and parameters stored on disk, and the long-term memory systems being developed for AI agents. Water is where knowledge accumulates and from which it can be drawn.在 AI 中，記憶與儲存即是水。這包括訓練資料庫、驅動檢索增強生成（RAG）的向量資料庫、儲存在硬碟上的權重與參數，以及為 AI 代理人開發的長期記憶系統。水是知識累積之處，也是知識汲取之源。

Water generates Wood — stored knowledge enables new data growth and new capabilities. The more an AI system can remember (context windows, persistent memory, episodic recall), the more effectively it can grow and adapt. And Water controls Fire — storage infrastructure constrains and channels computation. Without sufficient Water, Fire (computation) burns out of control, producing models that hallucinate because they lack the grounding of stored facts.水生木——儲存的知識推動新資料的成長與新能力。AI 系統能記住的內容越多（上下文視窗、持久記憶、情節回憶），就能越有效地成長與適應。而水剋火——儲存基礎設施限制並引導算力。沒有足夠的水，火（算力）就會失控燃燒，產生因缺乏儲存事實作為根據而產生幻覺的模型。

The race to build AI systems with longer context windows — from 4K tokens to 128K to 1M+ — is essentially the race to deepen Water. The deeper the reservoir, the more stable and reliable the system becomes.建構具備更長上下文視窗（從 4K Token 到 128K 再到 1M+）的 AI 系統競賽，本質上就是深化水之蓄量的競賽。水庫越深，系統就越穩定可靠。

A five-phase model with feedback in both directions — the structure is the argument.Figure 01具備雙向回饋的五行模型——結構即是論點。圖版 01

The Generating and Controlling Cycles in AIAI 中的相生與相剋循環

The power of the Five Elements framework lies not in the individual elements but in their relationships. The generating cycle (相生) and controlling cycle (相剋) create a self-regulating system.五行框架的力量不在於單一元素，而是在於它們之間的相互關係。相生循環（相生）與相剋循環（相剋）創造了一個自我調節的系統。

Cycle循環	Wu XingWu Xing	AI EquivalentAI 對應對象
Generating相生	Water → Wood水 → 木	Memory/storage produces new data capabilities記憶/儲存產生新資料能力
Generating相生	Wood → Fire木 → 火	Data fuels computation and training資料驅動計算與訓練
Generating相生	Fire → Earth火 → 土	Computation refines integration and architecture計算精煉整合與架構
Generating相生	Earth → Metal土 → 金	Integration enables algorithmic innovation整合推動演算法創新
Generating相生	Metal → Water金 → 水	Better algorithms improve memory efficiency更佳的演算法提升記憶體效率
Controlling相剋	Metal → Wood金 → 木	Algorithms determine how data is used演算法決定資料的使用方式
Controlling相剋	Wood → Earth木 → 土	Data quality constrains integration design資料品質限制整合設計
Controlling相剋	Earth → Water土 → 水	Architecture limits storage efficiency架構限制儲存效率
Controlling相剋	Water → Fire水 → 火	Storage capacity constrains computation scale儲存容量限制計算規模
Controlling相剋	Fire → Metal火 → 金	Compute budget constrains algorithm complexity算力預算限制了演算法的複雜度

Yin and Yang in AI SystemsAI 系統中的陰陽

Beyond the Five Elements, the fundamental duality of Yin Yang also maps onto AI in insightful ways:除了五行之外，陰陽的基本對立統一性也能以具啟發性的方式對應到 AI 上：

Yang (active, outward): Generation, inference, deployment, real-time interaction, creative output. The model actively producing text, images, or decisions.陽（主動、外向）：生成、推論、部署、即時互動、創意輸出。模型主動產生文字、圖片或決策。
Yin (receptive, inward): Training, data collection, evaluation, reflection, safety testing. The model absorbing information and being refined.陰（受容、內向）：訓練、資料收集、評估、反思、安全測試。模型吸收資訊並獲得提煉。

A healthy AI system needs both. A system that is all Yang — constantly generating without pause for evaluation — will produce confident nonsense. A system that is all Yin — endlessly training without deployment — never delivers value. The best AI products balance rapid iteration (Yang) with careful evaluation (Yin), just as a healthy person balances action with rest.一個健康的 AI 系統兩者都需要。全陽的系統——不斷生成而不停下來評估——會一本正經地胡說八道。全陰的系統——無休止地訓練而不部署——永遠無法提供價值。最優秀的 AI 產品會在快速疊代（陽）與審慎評估（陰）之間取得平衡，就像健康的人在行動與休息之間保持平衡一樣。

The emerging field of AI agents — systems that perceive, decide, and act autonomously — can also be understood through this lens. An agent that acts without reflection is excessive Yang. An agent that reflects without acting is excessive Yin. The most effective agents cycle between the two: observe (Yin), act (Yang), evaluate (Yin), adjust (Yang).AI 代理這個新興領域——能夠自主感知、決策與行動的系統——也可以透過這個視角來理解。只行動不反思的代理是陽氣過盛；只反思不行動的代理是陰氣過盛。最有效的代理會在兩者之間循環往復：觀察（陰）、行動（陽）、評估（陰）、調整（陽）。

Two modes that must keep trading places: the same balance problem, in different clothes.Figure 02必須不斷輪替的兩種模式：換湯不換藥的同一種平衡難題。圖版 02

Robotics and the Trigrams機器人學與八卦

Robotics adds a physical dimension to AI, and the Bagua (Eight Trigrams) offer an elegant way to categorize different types of robotic systems:機器人學為 AI 增添了實體維度，而八卦提供了一種優雅的方式來將不同類型的機器人系統進行分類：

Qian (☰ Heaven): Drones and aerial robots — upward movement, freedom, vision from above乾（☰ 天）：無人機與空中機器人——向上運動、自由、俯瞰視野
Kun (☷ Earth): Industrial robots and manufacturing arms — stability, repetition, grounded strength坤（☷ 地）：工業機器人與製造機械臂——穩定、重複、紮實的力量
Zhen (☳ Thunder): Search-and-rescue robots — sudden action, breaking through obstacles震（☳ 雷）：搜救機器人——迅捷行動、突破障礙
Xun (☴ Wind): Nanobots and soft robots — gentle, pervasive, penetrating small spaces巽（☴ 風）：奈米機器人與軟體機器人——溫和、無所不在、穿透狹小空間
Kan (☵ Water): Autonomous vehicles — flowing movement, adaptability, following paths坎（☵ 水）：自動駕駛車輛——順暢流動、適應力強、順應路徑
Li (☲ Fire): Computer vision systems — illumination, perception, seeing and identifying離（☲ 火）：電腦視覺系統——照亮、感知、辨識與看見
Gen (☶ Mountain): Stationary monitoring systems — stillness, vigilance, holding position艮（☶ 山）：靜止式監控系統——靜止、警惕、堅守陣地
Dui (☱ Lake): Social and companion robots — communication, joy, interaction with humans兌（☱ 澤）：社交與陪伴型機器人——溝通、喜悅、與人類的互動

Just as the Bagua represents all possible combinations of Yin and Yang across three dimensions, these eight categories cover the major paradigms of robotics. Real-world systems often combine multiple trigrams — an autonomous delivery robot is both Kan (flowing movement) and Dui (interacting with customers).正如八卦代表陰陽在三個維度上的所有可能組合，這八個類別也涵蓋了機器人技術的主要範式。現實世界中的系統往往融合了多個卦象——例如，一台自動配送機器人既具有坎卦（流動的運動）的特質，也具備兌卦（與客戶互動）的特質。

Note:注意： This framework is a philosophical lens, not a technical specification. The Five Elements and Bagua do not replace engineering analysis — they complement it by revealing patterns and imbalances that purely technical thinking might miss.這套框架是一副哲學眼鏡，不是技術規格。五行與八卦不會取代工程分析——它們是補充：揭示純技術思維可能看不見的模式與失衡。

The Imbalance Problem失衡問題

One of the most valuable applications of Wu Xing theory to AI is diagnosing imbalance. In traditional Chinese medicine, illness is understood as a disharmony among the Five Elements. The same diagnostic thinking can be applied to AI development:將五行理論應用於人工智慧最有價值的面向之一，在於診斷失衡。在中醫理論中，疾病被理解為五行之間的失調。這種診斷思維同樣可以應用於人工智慧的開發：

Excess Fire, deficient Water: Massive compute investment with inadequate data infrastructure. The model has processing power but lacks the knowledge depth to use it well. Symptom: hallucinations, overconfident errors.火過旺、水不足：海量算力投入，但資料基礎設施不足。模型擁有運算能力，卻缺乏充分利用該能力的知識深度。症狀：產生幻覺、過度自信的錯誤。
Excess Wood, deficient Metal: Data accumulation without algorithmic refinement. Mountains of data, but no efficient way to learn from it. Symptom: slow training, poor generalization.木過旺、金不足：資料堆積卻缺乏演算法精煉。有如山般的資料，卻缺乏從中有效學習的方法。症狀：訓練緩慢、泛化能力差。
Excess Metal, deficient Wood: Brilliant algorithms with insufficient data. The architecture is elegant but undernourished. Symptom: overfitting, poor real-world performance.金過旺、木不足：精妙的演算法卻缺乏足夠的資料。架構優雅卻營養不良。症狀：過擬合、現實世界表現不佳。
Deficient Earth: Powerful components that don't integrate. Great model, great data, great algorithms — but no deployment pipeline, no safety framework, no user interface. Symptom: "works in the lab, fails in production."土不足：強大的組件無法整合。優秀的模型、優質的資料、精良的演算法——但沒有部署流程、沒有安全框架、沒有使用者介面。症狀：「在實驗室可行，在生產環境失敗。」

The Daoist remedy is always the same: identify the imbalance, then strengthen the deficient element or moderate the excessive one through the generating and controlling cycles. This is not so different from what good AI engineering teams do — they just don't use this vocabulary.道家的解決之道始終如一：透過相生與相剋循環，識別失衡，然後補強不足的元素，或抑制過盛的元素。這與優秀的人工智慧工程團隊的做法並無二致——他們只是沒有使用這套詞彙而已。

AI Ethics Through Wu Xing透過五行看人工智慧倫理

The Five Elements framework also offers a natural language for AI ethics. Each element raises its own ethical question:五行框架也為人工智慧倫理提供了一種自然的思考語言。每個元素都提出了專屬的倫理思考：

Wood (Data): Where does the data come from? Who consented? Is the data diverse and representative, or overgrown with bias?木（資料）：資料來自何處？誰給予了授權？資料是否具備多樣性與代表性，還是充斥著偏見？
Fire (Computation): What is the environmental cost? Are we burning resources responsibly, or racing to build the biggest fire?火（運算）：環境成本為何？我們是在負責任地消耗資源，還是在盲目追求燃燒出最大的火焰？
Earth (Integration): Who is accountable when the system fails? Is there a human in the loop, or has Earth been neglected?土（整合）：當系統出錯時，誰該負責？過程中有真人介入監控，還是土元素被忽視了？
Metal (Algorithms): Are the algorithms transparent? Can we understand how decisions are made, or is the Metal opaque?金（演算法）：演算法是否透明？我們能否理解決策是如何做出的，抑或金元素晦澀不透明？
Water (Memory): What does the system remember about users? Is data retained with consent, or is the reservoir polluted?水（記憶）：系統上記錄了使用者的哪些資料？資料的保留是否取得同意，抑或蓄水池已被污染？

A truly balanced AI system addresses all five questions simultaneously. Focusing on only one or two — as most current AI ethics frameworks do — leaves the system vulnerable in the neglected dimensions.一個真正平衡的 AI 系統會同時處理這全部五個問題。像目前大多數 AI 倫理框架那樣僅關注其中一兩個問題，會使系統在被忽視的面向顯得脆弱。

What This Means for You這對您意味著什麼

Whether you are building AI systems or simply living in a world increasingly shaped by them, the Five Elements framework offers a way to think about technology that is both ancient and surprisingly relevant. It reminds us that:無論您是在建構 AI 系統，抑或只是生活在一個日益受其塑造的世界中，五行框架都提供了一種既古老又出奇切合現代科技的思考方式。它提醒我們：

Every system needs balance, not just power每個系統都需要平衡，而不僅僅是力量
Growth without structure becomes chaos缺乏結構的成長會演變成混亂
Efficiency (Metal) matters as much as scale (Fire)效率（金）與規模（火）同等重要
Memory (Water) is the foundation of intelligence記憶（水）是智慧的基石
Integration (Earth) is what turns raw capability into real value整合（土）能將原始能力轉化為真實價值

And just as your personal Bazi chart reveals which elements are strong and which need cultivation, understanding the "elemental composition" of any AI system can reveal where it excels and where it is vulnerable.正如您的個人八字命盤揭示了哪些五行強旺、哪些需要補益一樣，了解任何 AI 系統的「五行構成」，都能揭示其卓越之處與脆弱之處。

Discover Your Own Elemental Balance探索您專屬的五行平衡

Your Bazi chart reveals which of the Five Elements dominate your personality. Free, instant, no sign-up.您的八字命盤能揭示五行中哪種元素主導您的性格。免費、即時，無需註冊。

Reveal Chart開始測算