微軟揭秘人工智慧與區塊鏈融合，探索去中心化未來

微軟認識到區塊鏈技術的潛力及其與人工智慧（AI）的交叉點，表示有興趣優化現有技術，而不是創建自己的區塊鏈。微軟數位轉型、區塊鏈和雲端供應鏈總監約克·羅茲 (Yorke Rhodes) 表示，重點在於透過結合兩種技術的優勢來增強能力。專家預測，加密貨幣可能會在人工智慧的進步中發揮支援作用，特別是在去中心化圖形處理單元（GPU）和小語言模型等領域。儘管擔心去中心化網路中的延遲，但小組成員認為推理是可行的，並且可能會減少通訊開銷。預計監管機構將密切關注人工智慧的影響，但行業領導者表示，透過區塊鏈技術提高參與度和透明度可以解決監管問題。小組成員對人工智慧的未來表示樂觀，並展望了人工智慧在未來十年內被企業廣泛採用以及人工智慧（AGI）的潛力。

Microsoft and the Convergence of Artificial Intelligence and Blockchain Technology

微軟以及人工智慧和區塊鏈技術的融合

Intersection of Two Defining Technologies

兩種定義技術的交集

Microsoft, a leader in artificial intelligence (AI) research and investment, has expressed keen interest in the potential synergies between blockchain technology and AI. Speaking at the recent Cornell Blockchain Conference in New York, Yorke Rhodes, Microsoft's director for digital transformation, blockchain, and cloud supply chain, emphasized the transformative power of combining these two technologies.

人工智慧（AI）研究和投資領域的領導者微軟對區塊鏈技術與人工智慧之間的潛在協同效應表示了濃厚的興趣。微軟數位轉型、區塊鏈和雲端供應鏈總監約克·羅茲 (Yorke Rhodes) 最近在紐約舉行的康乃爾大學區塊鏈會議上發表講話，強調了這兩種技術相結合的變革力量。

"As these two technologies progress, we can create agents that bring together their collective capabilities. We're only scratching the surface," Rhodes stated.

「隨著這兩項技術的進步，我們可以創造出將它們的集體能力結合在一起的代理。我們只是觸及了表面，」羅茲說。

Alex Lin, moderator of the panel discussion titled "Crypto x AI," further probed Rhodes' views by inquiring whether Microsoft harbored aspirations of developing its own blockchain solution.

「加密 x AI」小組討論的主持人 Alex Lin 透過詢問微軟是否有開發自己的區塊鏈解決方案的願望，進一步探討了 Rhodes 的觀點。

Rhodes responded by acknowledging the significant advancements already underway in the crypto space, particularly within the open source community. "Why would we try to recreate something that has already attracted substantial investment?" he questioned. Consequently, Microsoft's current focus lies on optimizing existing technologies, such as layer-2 blockchain rollups.

羅德斯對此做出回應，承認加密貨幣領域已經取得了重大進展，特別是在開源社群內。 “為什麼我們要嘗試重建已經吸引了大量投資的東西？”他問。因此，微軟目前的重點在於優化現有技術，例如第 2 層區塊鏈匯總。

Crypto as the Enabler of Decentralized AI

加密貨幣作為去中心化人工智慧的推動者

Joining Rhodes and Lin on stage were Neil DeSilva, chief financial officer at PayPal Digital Currencies; Matt Stephenson, head of research at Pantera; and Jasper Zhang, CEO and co-founder at Hyperbolic Labs.

與羅茲和林一起上台的是 PayPal 數位貨幣財務長尼爾·德西爾瓦 (Neil DeSilva)；馬特‧史蒂芬森 (Matt Stephenson)，Pantera 研究主管；以及 Hyperbolic Labs 執行長兼共同創辦人 Jasper Zhang。

Stephenson expressed his belief that cryptocurrencies are well-positioned to serve as the underlying infrastructure for certain AI applications, particularly transformer and diffusion models. This is especially relevant given the anticipated prevalence of decentralized, multiagent AI systems in the future.

史蒂芬森表示，他相信加密貨幣完全可以作為某些人工智慧應用的底層基礎設施，特別是變壓器和擴散模型。考慮到未來去中心化、多智能體人工智慧系統的預期盛行，這一點尤其重要。

However, Stephenson recognized that crypto may play a supporting role to AI's dominant position. Rhodes concurred, acknowledging that a major trend like AI tends to dominate discussions at the expense of other emerging technologies, including crypto/blockchain and Web3.

然而，史蒂芬森意識到，加密貨幣可能對人工智慧的主導地位起到支撐作用。 Rhodes 對此表示同意，他承認像人工智慧這樣的主要趨勢往往會主導討論，而犧牲其他新興技術，包括加密/區塊鏈和 Web3。

Addressing Exaggerated Claims and Latency Issues

解決誇大的聲明和延遲問題

Lin raised concerns regarding exaggerated claims surrounding the intersection of blockchain networks and AI, emphasizing the difficulty in distinguishing hype from reality. He highlighted the example of decentralized graphics processing units (GPUs) and the often-overlooked issue of latency, the time it takes for data to transfer across a network.

林對圍繞區塊鏈網路和人工智慧交叉的誇大說法表示擔憂，並強調區分炒作與現實的困難。他強調了分散式圖形處理單元（GPU）的例子以及經常被忽視的延遲問題，即資料透過網路傳輸所需的時間。

Hyperbolic Labs' Zhang dismissed latency as a significant concern for decentralized networks like blockchains. He pointed out that centralized networks often suffer from lengthy data transfer times due to physical distances, while decentralized networks allow users to connect with nearby nodes for faster processing and reduced communication overhead.

Hyperbolic Labs 的張認為延遲是區塊鏈等去中心化網路的重要問題。他指出，由於物理距離的原因，中心化網路往往會遭受漫長的資料傳輸時間的困擾，而去中心化網路允許使用者與附近的節點連接，以實現更快的處理並減少通訊開銷。

Small Language Models and the Drive towards Edge AI

小語言模型與邊緣人工智慧的驅動力

Much attention in the AI field has focused on large language models (LLMs), but Rhodes emphasized the growing significance of smaller language models that operate efficiently on mobile devices and laptops. These smaller models enable a wider range of applications and take advantage of increased compute power at the edge.

AI 領域的大部分注意力都集中在大型語言模型 (LLM) 上，但 Rhodes 強調，在行動裝置和筆記型電腦上高效運行的小型語言模型的重要性與日俱增。這些較小的模型可實現更廣泛的應用，並利用邊緣運算能力的增強。

Microsoft has been actively developing small language AI models, including its Phi-3 family of open models, which are demonstrating capabilities approaching those of larger language models.

微軟一直在積極開發小語言人工智慧模型，包括其 Phi-3 系列開放模型，這些模型正在展示接近大型語言模型的功能。

Regulatory Landscape for AI and Crypto

人工智慧和加密貨幣的監管環境

Panelists anticipated increased regulatory scrutiny of AI in the coming years, similar to the attention given to cryptocurrencies. Lin expressed dissatisfaction with the U.S. regulatory approach, citing the heavy-handed stance of the Securities and Exchange Commission (SEC) towards crypto regulation.

小組成員預計未來幾年對人工智慧的監管審查將會加強，類似於對加密貨幣的關注。林表達了對美國監管方式的不滿，並列舉了美國證券交易委員會（SEC）對加密貨幣監管的嚴格立場。

DeSilva took a more nuanced view, acknowledging the potential frustrations of dealing with regulators but emphasizing their primary objective of protecting consumers from harm. He underscored the necessity of engaging with regulators to foster innovation that reaches a wider audience.

德席爾瓦的觀點更為細緻，他承認與監管機構打交道可能會遇到挫折，但強調監管機構的首要目標是保護消費者免受傷害。他強調了與監管機構合作的必要性，以促進影響更廣泛受眾的創新。

Blockchain as a Solution to AI's Opacity

區塊鏈作為人工智慧不透明性的解決方案

The opaque decision-making process of AI, known as the "black box" problem, poses challenges for regulators in mitigating potential consumer harm. Lin suggested that blockchain technology, with its transparency, immutability, and tracking capabilities, could provide a solution to this issue.

人工智慧的不透明決策過程（被稱為「黑盒子」問題）為監管機構減輕潛在消費者傷害帶來了挑戰。林認為，區塊鏈技術以其透明性、不變性和追蹤能力可以為這個問題提供解決方案。

Predictions for the Future of AI

對人工智慧未來的預測

In closing, panelists shared their visions for the future of AI. Zhang predicted widespread adoption of AI in the near future, with every company becoming an AI company. He expressed optimism that artificial generalized intelligence (AGI) could become a reality within the next five to 10 years, facilitated by decentralized infrastructure and increased compute power.

最後，小組成員分享了他們對人工智慧未來的願景。張預測，在不久的將來，人工智慧將被廣泛採用，每家公司都會成為人工智慧公司。他樂觀地認為，在去中心化基礎設施和運算能力增強的推動下，通用人工智慧（AGI）可能在未來五到十年內成為現實。

Rhodes forecasted the obsolescence of zero-knowledge proofs (ZK-proofs) in three years, to be replaced by fully homomorphic encryption (FHE), a technology that unlocks the value of data on untrusted domains without requiring decryption. FHE holds significant promise for addressing privacy concerns, particularly in sensitive areas such as healthcare.

羅茲預測零知識證明（ZK-proofs）將在三年內過時，取而代之的是完全同態加密（FHE），這是一種無需解密即可釋放不可信域資料價值的技術。 FHE 在解決隱私問題方面具有重大前景，特別是在醫療保健等敏感領域。

DeSilva emphasized the uncertain nature of concrete predictions in the technology industry. However, he expressed optimism for the eventual realization of AGI and its potential to benefit humanity through the combined efforts of all stakeholders.

德席爾瓦強調了科技業具體預測的不確定性。不過，他對AGI的最終實現及其透過所有利害關係人的共同努力造福人類的潛力表示樂觀。