應用商店
錢包

AI遇上Web3:2025年高科技大融合

AI遇上Web3:2025年高科技大融合

人工智能與區塊鏈技術於2025年正急速融合,為多個產業帶來新模式,有望徹底重塑數碼經濟。這種結合匯聚AI的強大運算能力,以及Web3的去中心化框架,不但解決了兩者的限制,亦為創新開拓更多可能性。資產管理公司Bitwise 預測 ,AI與加密貨幣結合最快到2030年可為全球GDP增加高達20萬億美元,展現此融合潛力巨大。


必知重點:

  • 自2022年以來,AI投資大幅飆升,如今美國創投資金42%流向AI公司,兩年前只有22%。
  • 以區塊鏈為基礎的Web3,為AI的「黑箱」問題帶來透明、不可更改的記錄解決方案。
  • 專家預期AI與加密貨幣結合可望在2030年前為全球GDP增添20萬億美元,但數據私隱及管治仍存挑戰。

shutterstock_2324952229.jpg

Web3的進化

由基本說起。

Web3代表互聯網科技的第三代,強調去中心化及用戶持有權,以區塊鏈基建為基礎,與過去互聯網時代大異其趣。1990年代的Web1.0是靜態只讀網站,Web2.0(2000-2010年代)則引入互動與社交媒體,但逐漸被掌握數據的大企業壟斷。

Web3這詞由以太坊聯合創辦人Gavin Wood於2014年提出,至2021年加密熱潮期間受關注。Web3奠基於開源區塊鏈網絡,以密碼學代幣啟動數碼資產擁有權及社群治理,讓無需中介的信任式交易及無需事前許可的創新成真。

Web3重要技術包括:比特幣、以太幣作P2P支付,以及在區塊鏈自動執行合約的智能合約。以太坊2015年引入智能合約,開創貨幣以外的應用,如去中心化金融(DeFi)、非同質化代幣(NFT)和去中心化自治組織(DAO)等。

2021年NFT藝術品高價成交、Facebook改名Meta掀起第一輪高潮,2022年市場回調後,發展步入理性階段。此期間,Web3基礎設施持續推進—以太坊升級、其他區塊鏈冒起、Layer-2加速交易速度等。

到2025年,生態圈已相當成熟。

關鍵教訓是:去中心化促進創意與全新商業模式,但用戶體驗、管治和安全仍需優化。現時區塊鏈已承載大量價值與數據,正需工具提升智能與易用性,AI正好進軍。

ChatGPT帶動AI速度大變革

AI成為高科技界新霸主。

ChatGPT於2022年底推出,象徵AI轉型時刻,被比喻為iPhone之於流動科技。短短兩年,生成式AI從小眾技術躍升為商業創新主軸。

到2024年初,超過75%受訪企業已在最少一項業務運用AI,65%持續用生成式AI,幾乎較去年倍增。

技術突破帶動此採用熱潮。OpenAI GPT-4等模型顯著提升AI生成內容能力,Google和Anthropic等對手亦相繼入場。硬件也變得關鍵,NVIDIA顯示卡需求爆炸,一度令公司市值破萬億美元。

傳統行業亦紛紛採納AI方案。金融機構用演算法檢測詐騙與管理投資組合,製造業利用AI機械人和預測維修,媒體公司用AI作內容個人化,政府部門則以AI提升公共服務。雲端運算令AI普及,模型可經API訪問或在租用伺服器上調校。

激增同時引來道德、私隱與可信度關注。帶有偏見的演算法或失靈聊天機械人多次成為新聞熱話,推動監管措施出台。歐盟擬定AI法規作指引,部分地區甚至因私隱暫禁某些AI應用。

AI與Web3相輔相成

2025年AI與Web3結合展現強大協同效應,雖然一者重中央數據,一者強調去中心化。AI大大提升區塊鏈應用的易用性和智能化。區塊鏈過去缺乏友善介面,只能執行簡單邏輯,加入AI後卻可締造智能化合約及反應更靈敏的服務。

AI可分析實時數據以觸發智能合約,令合約根據複雜情況前先處理資訊再執行。在去中心化金融領域,如自動貸款合約可據市場狀況或信貸評級自動調整條款。AI界面可引導用戶完成區塊鏈操作,將意圖轉成交易及個人化建議—消除以往令加密貨幣主流化受限制的易用門檻。

反之,Web3亦能彌補AI弱點。區塊鏈的透明度可破解AI「黑箱」問題,把訓練數據、參數及決策過程存於不可更改帳本,產生可查證紀錄。監管機構及用戶可檢視AI訓練過程,核實系統可信度。基於區塊鏈的身份系統亦可為AI代理添加認證,建立有紀錄的數碼身份,尤其重要因自主AI逐步代表人類處理交易。

Web3對資料擁有權的新做法,能取代中心化數據庫。個人可選擇提供數據訓練AI,同時保留控制權及獲得代幣回報,而非讓科技公司主導。

儘管結合潛力巨大,依然面臨挑戰。AI需要龐大數據時,區塊鏈卻公開透明,私隱問題嚴峻。技術如聯邦學習或零知識證明,可望讓AI運作時不用暴露敏感資料,惟尚在發展中。規管層面,如何令GDPR等法規配合資料一經寫入區塊鏈就不能刪除的特性,是大難題。

AI在Web3的真實應用

金融服務革新

去中心化金融(DeFi)是AI與Web3結合最具潛力的領域之一。2025年,AI正讓DeFi更精明易用,包括信用風險評估、收益策略最佳化、自主執行交易等。

機械顧問全天候監察加密市場,按用戶設定與風險承受能力調整資產組合。這些AI代理猶如透明上鏈的小型對沖基金,把複雜金融策略普及予小型投資者。

區塊鏈支付同樣受惠AI。穩定幣—與法定貨幣掛勾的加密貨幣—由2020年流通量40億美元暴升至2024年底近2000億美元。當AI應用於穩定幣網絡後,複雜財務操作可自動化企業資金流,AI按市場動態觸發適當支付或對沖,務求令常規流程自動發生,節省人手並提升效率。

AI推動Web3平台誕生嶄新金融產品。參數化保險(parametric insurance)—只要某些狀況出現就自動賠償—能結合AI即時分析外部數據。如此一來,偏遠地區農民可購入微型氣候保險,由AI探測旱災觸發穩定幣賠償,無須任何文件。

實際情況:

AI已被整合入去中心化金融(DeFi)平台,例如 Circle's USDC穩定幣,實現即時、AI驅動的穩定幣交易與智能資產管理。Aave及MakerDAO等項目同樣運用AI加強鏈上借貸、交易及風險評估能力。

去中心化治理進化

去中心化自治組織(DAO)正應用AI提升協作與決策效率。以往DAO管治人數眾多,討論與投票常陷混亂。AI可分析社交平台情緒、在正式投票前整合輿論,並將冗長討論濃縮成重點摘要,降低參與障礙。 barriers.

AI agents themselves are becoming participants in DAO ecosystems. Experiments include AI agents receiving grants to develop investment strategies, essentially functioning as fund managers under DAO oversight. In other cases, bots handle routine tasks like treasury rebalancing or community moderation according to guidelines established by human members.

AI 代理本身亦開始成為 DAO 生態系統內的參與者。有些實驗給予 AI 代理資助,讓其發展投資策略,基本上等同於由 DAO 監督下的基金經理。另一啲情況下,機械人會依照人類成員訂立既指引處理日常事務,例如財庫再平衡或者社群管理。

Treasury management represents a concrete application where AI demonstrates value. Many DAOs control significant funds, sometimes exceeding $100 million in crypto assets. AI-based portfolio management tools can automatically diversify assets or generate yield through DeFi protocols while adhering to community-defined risk parameters.

財庫管理係一個 AI 可以實際發揮價值嘅應用場景。好多 DAO 管理住相當大嘅資金,有時甚至超過一億美金嘅加密資產。基於 AI 嘅投資組合管理工具可以自動多元化資產,或者透過 DeFi 協議產生收益,並且遵從社群所訂既風險參數。

These agents follow encoded rules with all transactions logged on-chain, providing complete transparency.

呢啲代理係跟從已編碼嘅規則行事,所有交易都會記錄喺鏈上,確保完全透明。

Organizations approach AI integration cautiously, typically keeping humans in supervisory roles. Trust develops by allowing AI to execute strategies while humans retain policy-setting authority and override capabilities. Web3's transparency makes AI actions traceable in ways traditional corporate AI often isn't—every on-chain action by a DAO's AI can be audited by members in real-time.

各組織對於整合 AI 採取審慎態度,通常會保留人類負責監督。建立信任既方法係讓 AI 執行策略同時由人類保留制定政策同覆核決定權。Web3 嘅透明度令 AI 行為變得可以追蹤,而以往傳統企業 AI 做唔到—DAO 裏面 AI 嘅任何鏈上操作,會員都可即時審核。

In the real world:

Decentralized Autonomous Organizations (DAOs), like Aragon and Compound, are increasingly employing AI tools for treasury management, governance analytics, and community moderation. Notably, DAOstack has experimented with AI-driven sentiment analysis and automated decision-making to streamline governance processes and improve organizational efficiency.

現實中,好似 Aragon 同 Compound 呢啲分散式自治組織(DAO),越來越多應用 AI 工具用作財庫管理、治理分析同社群管理。DAOstack 更曾經試過用 AI 驅動既情緒分析同自動決策,以簡化治理流程同提升組織效率。

Creative Economy Innovations

The creative economy built around Web3 is undergoing transformation through AI integration. Artists and developers increasingly use AI tools to generate content that is owned, traded, or experienced on blockchain networks. This spans visual art, profile-picture collections, music, literary works, and metaverse environments.

Web3 周邊既創意經濟正因為 AI 整合而出現變革。藝術家同開發者愈來愈多咁用 AI 工具創造內容,呢啲內容可以擁有、交易或者喺區塊鏈網絡上體驗。範疇覆蓋視覺藝術、大頭貼系列、音樂、文學作品、甚至元宇宙環境。

Generative art NFTs represent a notable manifestation. Artists train AI models on specific styles or concepts, producing endless variations that can be minted as unique tokens.

生成藝術 NFT 就係一個最明顯嘅例子。藝術家用 AI 模型訓練特定既風格或概念,創造出無盡變化,而每個都可以鑄造成獨一無二嘅代幣。

Major auction houses have validated this trend, with Christie's holding its first auction dedicated to AI-generated art in early 2025, achieving over $700,000 in sales despite mixed results.

大型拍賣行已經認證咗呢個潮流,其中佳士得喺 2025 年初舉行第一場專為 AI 創作藝術品而設既拍賣,雖然競標結果參差,但總成交都超過七十萬美金。

Interactive NFTs are emerging with embedded AI functionality. Examples include virtual pets or avatars with AI personalities that owners can interact with, evolving over time. This makes NFTs dynamic experiences rather than static collectibles. Web3 games similarly incorporate AI to create more realistic non-player characters capable of improvising dialogue and adapting to player actions.

帶有 AI 功能嘅互動型 NFT 正慢慢出現。例如有啲係 AI 性格既虛擬寵物或分身,持有人可以互動,而呢啲虛擬角色仲會隨住時間成長。咁樣就令 NFT 由一件死物變成一個動態體驗。Web3 遊戲都一樣,引入 AI 打造更逼真、有自主對話能力並會根據玩家行動改變既 NPC。

AI-generated content marketplaces are developing on Web3 platforms, allowing creators to mint AI-generated music as NFTs with automatic royalty distribution to both model creators and musicians. Some DAOs commission AI models to generate intellectual property that community members collectively manage and license across media formats, with revenue shared through tokens.

Web3 平台上已經開始出現 AI 內容市場。創作者可以將 AI 生成音樂鑄造成 NFT,自動分發版稅俾模型開發者同音樂人。有啲 DAO 甚至會委託 AI 模型產出智慧財產俾社群集體管理和授權至唔同媒體格式,而收入則以代幣分配。

The boundaries between creator, tool, and owner are blurring in fascinating ways. Web3 can record contributions to creative works and use smart contracts to allocate appropriate revenue shares. This potentially addresses controversies around AI art by automatically compensating artists whose styles influence AI outputs.

創作者、工具同擁有人嘅界線正以有趣方式模糊化。Web3 技術可以紀錄對創作作品嘅各項貢獻,再用智能合約自動分配收益。呢種方式有機會解決 AI 藝術爭議,例如自動補償影響咗 AI 產出既藝術家。

In the real world:

AI-generated art is making waves in the NFT market, highlighted by Christie’s first dedicated AI art auction featuring artists like Refik Anadol and platforms like Art Blocks. Interactive NFT projects, including Altered State Machine (ASM), are embedding AI into NFTs, allowing dynamic interactions and evolving digital collectibles.

現實中,AI 生成藝術係 NFT 市場掀起浪潮,佳士得首次 AI 藝術專場集合咗 Refik Anadol 等藝術家以及 Art Blocks 等平台。互動 NFT 項目如 Altered State Machine (ASM) 將 AI 加入 NFT,令收藏品可以動態互動和進化。

Gaming Ecosystem Advancement

Web3 gaming is experiencing significant enhancement through AI integration, with improvements both within game worlds and in development processes. Inside games, AI powers non-player characters and content generation, creating richer experiences. Characters in blockchain games can remember player interactions and evolve over time, with memories stored as data attached to NFTs, creating personalized gameplay narratives.

Web3 遊戲生態因整合 AI 而有大幅提升,無論遊戲內容定開發流程都得到改進。遊戲內,AI 推動住 NPC 角色同內容產生,帶來更豐富體驗。區塊鏈遊戲角色可以記住玩家互動並隨時間成長,呢啲記憶會以數據方式附喺 NFT,帶來個人化既遊戲故事。

Game studios utilize generative AI for procedural content creation, rapidly producing diverse landscapes, items, and dialogue. Industry-standard game engines now include built-in AI tools for generating textures and simulating physics, helping Web3 games achieve visual and narrative depth comparable to mainstream titles.

遊戲開發工作室用生成式 AI 去做程序內容生成,好快咁造出多元環境、裝備同對白。行業標準既遊戲引擎而家都內置 AI 工具,生成材質、模擬物理效果,幫 Web3 遊戲做到主流遊戲咁有層次感。

AI is dramatically reducing development time and costs for blockchain games. According to industry leaders, AI-assisted development—generating code snippets, designing artwork, testing for bugs—has cut production timelines by approximately 65% over the past year. This enables smaller studios to compete effectively by using AI for labor-intensive tasks like character animation or economy balancing. One mobile developer described using AI to simulate thousands of player strategies overnight to optimize token reward systems, work that would traditionally require extensive testing teams.

AI 大幅減少咗區塊鏈遊戲既開發周期同成本。有行內人指,AI 自動生成程式碼、設計美術、檢測錯誤等,過去一年令生產時間縮短咗約65%。咁一來,細 studio 只要用 AI 做角色動畫或經濟系統平衡等繁重工序都可以有效競爭。有個手機開發商仲分享過,點用 AI 晚上一夜之間模擬幾千個玩家策略,優化代幣獎勵系統,呢啲以往要用成隊 QA 測試人馬去做。

AI is also improving economic systems within play-to-earn games. Balancing economies where players earn real value presents complex challenges—AI modeling predicts how virtual economies respond to changes by analyzing player data, helping designers maintain stability.

AI 同時優化緊賺錢型(play-to-earn)遊戲經濟系統。呢類玩家會賺到真實價值,平衡經濟難度極高—AI 模型會透過分析玩家數據,預測虛擬經濟對變動既反應,幫助設計師維持穩定。

AI can personalize financial experiences, offering newer players accessible quests with reasonable rewards while directing veterans toward community events that sustain engagement.

AI 亦都可以令金融體驗個人化,新手玩家會獲得容易入手且好合理嘅任務同獎勵,而老手則被引導參加大型社群活動繼續參與。

In the real world:

Web3 gaming platforms such as Illuvium and Immutable are leveraging AI to enhance gameplay with adaptive NPCs and procedurally generated content. Axie Infinity and upcoming blockchain-based games from studios using Unreal Engine 5 integrate advanced AI tools for richer, more personalized player experiences.

現實中,Illuvium、Immutable 等 Web3 遊戲平台正運用 AI 增強遊戲性,有自適應 NPC 啦、程序內容生成等等。Axie Infinity 以及用 Unreal Engine 5 開發既新一代區塊鏈遊戲都整合咗先進 AI 工具,提升玩家體驗至更豐富、更個人化既層次。

Infrastructure and Security Developments

Behind-the-scenes infrastructure represents a foundational area where AI and Web3 are converging. This includes enhancing blockchain networks and using Web3 principles to decentralize AI development itself. Computing power illustrates this synergy. AI model training requires immense computational resources, traditionally limited to major tech companies. Meanwhile, cryptocurrency mining has created globally distributed high-powered computer networks that are often underutilized.

幕後既基礎設施正係 AI 同 Web3 結合既關鍵領域,涵蓋增強區塊鏈網絡,以及用 Web3 原則推動 AI 發展去中心化化。例如運算資源方面,AI 模型訓練需要海量運算力,過往只有大科企先有。另一方面,加密幣挖礦創造咗分布全球既強大運算網絡,但經常用唔晒。

Decentralized compute marketplaces have emerged to bridge this gap. Networks allow crypto miners and data centers to rent excess GPU capacity to AI researchers, with blockchain-based systems handling payments. This creates distributed "supercomputers" without reliance on single providers, aligning with Web3's anti-monopoly philosophy while offering miners alternative revenue streams.

去中心化運算市場已經開始出現,橋接返呢個缺口。呢啲網絡讓礦工或數據中心出租閒置 GPU 俾 AI 研究員,付款用區塊鏈處理。咁就可以組建分散式「超級電腦」,唔需要依賴單一供應商,又符合 Web3 反壟斷精神,亦為礦工帶來新收入來源。

Similar decentralization is occurring with datasets. Web3 data marketplaces enable providers to sell access to datasets for AI training, with all transactions recorded on blockchain. This creates auditable trails showing which data trained specific AI models, addressing transparency concerns. Several organizations are exploring "model provenance" on blockchain, where each AI model update is recorded like a software repository commit.

資料集方面都開始去中心化。Web3 數據市場容許供應商賣數據用作 AI 訓練,交易全部記錄喺區塊鏈。可以追蹤到底邊個數據訓練過咩模型,對提升透明度有好大幫助。有機構仲試緊將「模型來源」記錄喺鏈上,每次 AI 模型更新都似軟件代碼 commit 咁留底。

Security within crypto infrastructure benefits from AI deployment. The anonymous, irreversible nature of blockchain transactions has attracted fraudulent activity that traditional monitoring struggles to detect. Exchanges and protocols employ machine learning models to analyze transactions in real-time, flagging anomalies and suspicious patterns. These systems can identify potential account compromises or prevent attacks like flash loans by simulating transaction impacts before execution.

加密基礎設施保安方面,AI 部署都有好大作用。區塊鏈天生匿名、唔可逆好吸引欺詐活動,傳統監控系統難以偵測。交易所同協議往往用機器學習模型即時分析交易,搵出異常同可疑模式,可以識別帳戶被盜、甚至預先模擬交易過程防止閃電貸攻擊一類事故。

Blockchain is similarly securing AI systems. As models become valuable intellectual property, verifying their integrity becomes crucial. Blockchain can timestamp and hash model parameters, effectively creating tamper-evident fingerprints. This has spawned experimental "AI model NFTs" representing ownership of specific AI versions, potentially including smart contracts that automatically compensate original creators through royalties.

區塊鏈同樣幫到 AI 系統保安。模型愈來愈有價值,確保完整性就更重要。區塊鏈可以為模型參數加上時間戳及雜湊值,變成防篡改指紋。因應咁樣,有啲實驗性「AI 模型 NFT」出現,代表指定 AI 版本既擁有權,甚至可以包埋智能合約定期支付版稅俾原創人。

In the real world:

Projects like Render Network, Bittensor, and Ocean Protocol exemplify decentralized marketplaces providing GPU computing power and AI data-sharing services on blockchain. Meanwhile, exchanges including Binance employ machine learning to bolster blockchain security, fraud detection, and infrastructure resilience, enhancing user protection across crypto ecosystems.

現實中,Render Network、Bittensor、Ocean Protocol 等項目就係好例子,提供 GPU 運算同 AI 數據共享既去中心化區塊鏈市場。至於 Binance 等交易所,亦用機器學習強化區塊鏈保安、防詐及基礎設施韌性,提升加密生態用戶安全。

The Future of AI-Web3 Convergence

As the AI-Web3 intersection progresses through 2025, early hype is transitioning toward practical implementation. The use cases examined demonstrate tangible progress across finance, governance, creativity, gaming, and infrastructure.

隨住 AI 及 Web3 交匯點發展去到 2025,初期既炒作逐步轉向實際落地應用。以上各個案例都展現出金融、治理、創意、遊戲以至基礎設施等方面已有明顯進展。

Institutional involvement is shaping developmental trajectories. Financial organizations initially cautious about both technologies are exploring combined applications for longstanding problems. Consulting firms advise clients on integrated strategies for supply chains and identity management. Even governments are utilizing blockchain to secure public data for AI analysis. Regulatory approaches are becoming more holistic, recognizing that AI-Web3 applications span multiple domains simultaneously.

機構參與逐步影響住未來發展路向。原本對新技術很保守既金融機構都開始嘗試用兩者合力解老問題。顧問公司已經為客戶設計結合供應鏈同身份管理既策略。有啲政府則用區塊鏈保障公共數據俾 AI 分析。監管取態愈來愈全面,因為 AI-Web3 應用同時覆蓋多個領域。

Industry standards and collaborations are emerging at this intersection. Technical communities that historically operated separately are increasingly combining expertise, with interdisciplinary research exploring topics like blockchain incentives for federated learning or AI-optimized consensus algorithms.

呢個交匯點都開始出現行業標準同協作。過去各自為政既技術社群而家愈來愈多合作,有跨界研究項目討論區塊鏈如何獎勵聯邦學習,點樣用 AI 優化共識演算法等等。

Looking ahead 3-5 years, several scenarios appear plausible. Consumer applications combining Web3 and AI might achieve mainstream adoption, perhaps as personal

(原文未完,如需繼續翻譯,歡迎補充補全內容。)assistants 管理數碼資產和身份,同時保留數據擁有權。企業採用這些技術後,全球供應鏈的重要部分有機會透過區塊鏈記錄及由人工智能系統優化。金融基礎設施有可能透過人工智能整合,把中央銀行數碼貨幣和去中心化金融結合起來。

免責聲明及風險提示: 本文資訊僅供教育與參考之用,並基於作者意見,並不構成金融、投資、法律或稅務建議。 加密貨幣資產具高度波動性並伴隨高風險,可能導致投資大幅虧損或全部損失,並非適合所有投資者。 文章內容僅代表作者觀點,不代表 Yellow、創辦人或管理層立場。 投資前請務必自行徹底研究(D.Y.O.R.),並諮詢持牌金融專業人士。
AI遇上Web3:2025年高科技大融合 | Yellow.com