Meta's AI Open-Source Claims Draw Criticism, OSI Pushes For Transparency
Meta's AI Open-Source Claims Draw Criticism, OSI Pushes For Transparency
Meta Platforms Inc. (NASDAQ:META) received criticism from the Open Source Initiative (OSI) for using the term "open-source" to describe its AI models.
meta platforms公司（納斯達克：META）因將其人工智能模型描述爲「開源」而受到開源倡議組織（OSI）的批評。
The criticism focuses on the company's Llama family of large language models, with OSI head Stefano Maffulli accusing Meta of misleading users and "polluting" the concept of open-source technology.
這一批評集中在該公司的大型語言模型Llama系列上，OSI主席斯特凡諾·馬菲利指責Meta誤導用戶，並「污染」了開源技術的概念。
Maffulli's comments in an interview with the Financial Times highlight concerns about Meta's claims when the European Commission and other bodies push for accurate open-source technology beyond any company's control.
馬菲利在接受《金融時報》採訪時發表的評論凸顯了對Meta聲稱的擔憂，當歐洲委員會和其他組織推動精準的開源技術超出任何公司控制範圍時。
Also Read: Alibaba And Tencent Lap Up Meta's AI Large Language Model
閱讀更多：阿里巴巴和騰訊搶先獲得Meta的人工智能大型語言模型
According to Maffulli, this confusion could undermine long-term efforts to create user-controlled AI systems rather than be dominated by a few tech giants.
根據馬菲利的說法，這種混淆可能會破壞建立用戶控制的人工智能系統的長期努力，而不是被少數科技巨頭所主導。
Meta has positioned Llama as a significant player in the AI space, claiming the model has been downloaded more than 400 million times. It's part of a broader wave of AI models challenging the proprietary systems of companies like Microsoft Corp (NASDAQ:MSFT) backed OpenAI and Alphabet Inc (NASDAQ:GOOGL) (NASDAQ:GOOG) Google.
Meta將Llama定位爲人工智能領域的重要參與者，聲稱該模型已被下載超過 40000萬 次。它是一波挑戰微軟公司（納斯達克：MSFT）支持的OpenAI和Alphabet Inc（納斯達克：GOOGL）（納斯達克：GOOG）Google等公司專有系統的AI模型浪潮的一部分。
However, Maffulli points out that Llama, like many other "open-source" models, does not provide the full transparency or flexibility that traditional open-source software allows.
然而，馬菲利指出，與許多其他「開源」模型一樣，Llama不能提供傳統開源軟件允許的完全透明性或靈活性。
While Meta allows developers to access the model's weights, it restricts complete experimentation by keeping the development process and algorithms secret.
雖然Meta允許開發人員訪問模型的權重，但通過保持開發過程和算法的祕密，它限制了完整的實驗。
At SIGGRAPH 2024, Meta chief Mark Zuckerberg emphasized his frustration with closed-source AI platforms, criticizing them for stifling innovation.
在SIGGRAPH 2024，meta chief馬克·扎克伯格強調了他對閉源AI平台的沮喪，批評它們扼殺創新。
He strongly preferred open-source models, arguing they allow for more creative freedom and development. Zuckerberg's stance aligns with his push for Meta's Llama 3.1, the company's largest open-source AI model.
他強烈偏愛開源模型，認爲它們能夠提供更多創造自由和發展空間。扎克伯格的立場與他推動Meta Llama 3.1的努力保持一致，這是公司最大的開源AI模型。
He also pointed out the irony of companies like OpenAI, which, despite their name, lead the development of closed-source AI models.
他還指出了像OpenAI這樣的公司的諷刺之處，儘管他們的名稱是「開放AI」，他們還是在開發閉源AI模型。
Nvidia Corp (NASDAQ:NVDA) recently introduced an open-source AI model. The new NVLM 1.0 family of large multimodal language models enhances text capabilities, vision, and language tasks.
Nvidia公司（納斯達克股票代碼：NVDA）最近推出了一款開源AI模型。新的NVLm 1.0系列大型多模式語言模型增強了文本能力、視覺和語言任務。
Additionally, Nvidia unveiled Unreal Engine 5 on-device plugins for its Ace platform, enabling AI-powered MetaHuman characters and new tools for developers to create AI-driven facial animations on Autodesk Maya.
此外，英偉達還爲其Ace平台推出了Unreal Engine 5的設備插件，實現了基於AI的MetaHuman角色和爲開發者創建在Autodesk Maya上進行AI驅動面部動畫的新工具。
Price Action: META stock is up 0.63% at $580.40 premarket at last check Thursday.
股價表現：meta股票漲幅爲0.63%，報580.40美元，在週四最後一次查看時爲盤前交易。
Also Read:
還閱讀：
- Amazon Prime Video India to Introduce Ads in 2025, Intensifying Rivalry with Reliance and Disney+
- 亞馬遜Prime Video印度將於2025年推出廣告，加劇與Reliance和迪士尼爭端。
Image via Shutterstock
圖片來自shutterstock。
譯文內容由第三人軟體翻譯。
