Gemini (語言模型)
開發者 | Google DeepMind Google AI |
---|---|
初始版本 | 2023年12月6號 |
話 | English |
軟件類別 | Large language model |
許可協議 | Proprietary |
網站 | gemini |
Google Gemini 係一系列由 Google DeepMind 發展嘅多模態學習 大型語言模型,係 LaMDA 同 PaLM 2 嘅繼承者。家族成員包括 Gemini Ultra、Gemini Pro、Gemini Flash 同 Gemini Nano,喺 2023 年 12 月 6 日公佈,定位為OpenAI 嘅 GPT-4 嘅競爭對手。佢為同名聊天機械人提供動力。
歷史
[編輯]發展
[編輯]Google 喺 2023 年 5 月 10 日嘅 Google I/O 主題演講中宣佈由子公司 Google DeepMind 發展嘅 大型語言模型 (LLM) Gemini。佢被定位為 PaLM 2 嘅更強繼承者,喺同場活動中亦有公佈,Google CEO Sundar Pichai 表示 Gemini 仍喺早期發展階段。[1][2] 與其他 LLM 不同,Gemini 唔係單靠文本語料庫訓練,而係設計成多模態學習,即可以同時處理多種類型數據,包括文本、圖片、音頻、視頻同源代碼。[3] 佢係 DeepMind 同 Google Brain 合作發展嘅,兩者喺上個月合併為 Google DeepMind。[4] 喺《Wired》嘅訪問中,DeepMind CEO Demis Hassabis 吹捧 Gemini 嘅先進功能,佢相信呢個算法可以超越 OpenAI 嘅 ChatGPT,後者運行喺 GPT-4 上,並且憑借其日益受歡迎而被 Google 以 LaMDA 同 Bard 積極挑戰。Hassabis 強調 DeepMind 嘅 AlphaGo 計劃嘅優勢,2016 年當佢擊敗 圍棋冠軍 李世乭 時,引起全球關注,表示 Gemini 將結合 AlphaGo 同其他 Google–DeepMind LLMs 嘅力量。[5]
2023 年 8 月,《The Information》發佈一份報告,概述 Google 嘅 Gemini 路線圖,透露公司目標喺 2023 年底推出。根據報告,Google 希望通過結合大多數 LLMs 中嘅對話文本能力同 人工智能 驅動嘅圖像生成,超越 OpenAI 同其他競爭對手,使佢能夠創建上下文圖片,並適應更廣泛嘅使用情景。[6] 同 Bard 一樣,[7] Google 聯合創辦人 Sergey Brin 被召回出山協助發展 Gemini,同時仲有數百名來自 Google Brain 同 DeepMind 嘅工程師;[6][8] 佢後來被認為係 Gemini 嘅「核心貢獻者」。[9] 因為 Gemini 喺訓練過程中使用咗 YouTube 視頻嘅文字記錄,律師被引入過濾出任何潛在受版權保護嘅材料。[6]
隨著 Gemini 即將推出嘅消息,OpenAI 加快咗其將 GPT-4 同類似 Gemini 嘅多模態功能集成嘅工作。[10] 《The Information》喺九月報導幾家公司已經獲得「早期版本」LLM 嘅早期訪問權限,Google 計劃通過 Google Cloud 嘅 Vertex AI 服務提供畀客戶。該刊物亦指出,Google 正準備 Gemini 同 GPT-4 及 Microsoft 嘅 GitHub Copilot 競爭。[11][12]
發佈
[編輯]2023 年 12 月 6 日,Pichai 同 Hassabis 喺一場虛擬新聞發佈會上宣佈「Gemini 1.0」。[13][14] 佢包括三個模型:Gemini Ultra,設計用於「高度複雜嘅任務」;Gemini Pro,設計用於「廣泛嘅任務」;同 Gemini Nano,設計用於「設備上嘅任務」。喺發佈時,Gemini Pro 同 Nano 被整合到 Bard 同 Pixel 8 Pro 智能手機中,而 Gemini Ultra 將為「Bard Advanced」提供動力,並喺 2024 年初提供畀軟件開發者。Google 計劃將 Gemini 整合到嘅其他產品包括 搜尋、廣告、Chrome、Google Workspace 嘅 Duet AI 同 AlphaCode 2。[15][14] 佢目前僅支持英文。[14][16] Google 宣佈 Gemini 係其「最大同最強大嘅 AI 模型」,設計模仿人類行為,[17][14][18] 公司表示,由於需要「廣泛嘅安全測試」,Gemini 唔會喺翌年之前廣泛供應。[13] Gemini 喺 Google 嘅 張量處理單元 (TPUs) 上進行訓練同提供動力,[13][16] 呢個名稱係參考 DeepMind–Google Brain 合併同 NASA 嘅 雙子座計劃。[19]
Gemini Ultra 被認為喺多個行業基準測試中表現優於 GPT-4、Anthropic 嘅 Claude 2、Inflection AI 嘅 Inflection-2、Meta 嘅 LLaMA 2 同 xAI 嘅 Grok 1,[20][21] 而 Gemini Pro 被認為優於 GPT-3.5。[22] Gemini Ultra 亦係首個喺 57 科目 大規模多任務語言理解 (MMLU) 測試中超越人類專家嘅語言模型,獲得 90% 嘅分數。[22][23] Gemini Pro 喺 12 月 13 日向 Google Cloud 客戶嘅 AI Studio 同 Vertex AI 提供,而 Gemini Nano 亦將提供畀 Android 開發者。[24][25]
架構
[編輯]Google Gemini 係一個多模態學習模型,即能夠同時處理文本、圖片、音頻、視頻同源代碼。[22] 根據 Google,佢包含多層深度學習神經網絡,可以模仿人類嘅自然語言理解同生成能力。佢嘅核心架構係基於Transformer,一種被證明非常有效嘅深度學習模型架構。
應用
[編輯]Google 計劃將 Gemini 整合到其多個產品中,包括搜尋、廣告、Chrome、Google Workspace 嘅 Duet AI 同 AlphaCode 2。[26][27] Gemini 嘅多模態能力亦使佢適合用於圖像生成、視頻理解同編碼建議等各種應用。Google 表示,佢哋會繼續優化 Gemini 嘅性能同擴展其應用範圍。
競爭
[編輯]作為一個先進嘅大型語言模型,Gemini 面臨來自多個競爭對手嘅挑戰,包括 OpenAI 嘅 GPT-4、Anthropic 嘅 Claude 2、Inflection AI 嘅 Inflection-2、Meta 嘅 LLaMA 2 同 xAI 嘅 Grok 1。[20][21] 然而,Google 對 Gemini 嘅前景充滿信心,相信佢嘅多模態能力同強大性能可以使佢喺市場上脫穎而出。
更新
[編輯]Google 喺 2024 年 1 月同 三星 合作,將 Gemini Nano 同 Gemini Pro 整合到其 Galaxy S24 智能手機系列中。[28][29] 喺嗰個月嘅之後,Bard 同 Duet AI 喺 Gemini 品牌下統一,[30][31] 並喺 Google One 訂閱服務中推出新嘅 "AI Premium" 層級,帶嚟 "Gemini Advanced with Ultra 1.0"。[32] Gemini Pro 亦都進行全球發布。[33]
到咗 2 月,Google 喺有限容量嘅情況下推出 "Gemini 1.5",呢個版本比 1.0 Ultra 更加強大同高效。[34][35][36] 呢個 "step change" 係通過多種技術進步實現嘅,包括新嘅架構、mixture-of-experts 方法,同埋一個更大嘅一百萬個 token 嘅context window,呢個大約相當於一個鐘嘅靜音視頻、11 個鐘嘅音頻、30,000 行代碼或者 700,000 字。[37] 喺同一個月,Google 推出咗 Gemma,一個家族嘅 free and open-source LLMs,作為 Gemini 嘅輕量版本。佢哋有兩個尺寸,分別係包含兩億同七億參數嘅神經網絡。多個出版物視呢個作為回應 Meta 同其他人開源佢哋嘅 AI 模型,以及與 Google 一貫保持 AI 專有權嘅做法形成鮮明對比嘅重大轉變。[38][39][40]
喺 Google I/O 2024 上,Gemini 1.5 Flash 發布咗。[41]
技術規格
[編輯]第一代嘅 Gemini("Gemini 1")有三個模型,佢哋有相同嘅軟件架構。佢哋係只解碼嘅transformers,經過修改可以高效喺 TPU 上進行訓練同推理。佢哋有 32,768 個 token 嘅上下文長度,採用多重查詢注意力。兩個版本嘅 Gemini Nano,Nano-1(18 億參數)同 Nano-2(32.5 億參數),係由更大嘅 Gemini 模型蒸餾出嚟,設計畀邊緣設備例如智能手機使用。由於 Gemini 係多模態嘅,每個上下文窗口可以包含多種形式嘅輸入。唔同嘅模式可以交錯出現,唔需要按固定嘅順序展示,從而實現多模態對話。例如,用戶可以用混合文本、圖片、視頻同音頻開啟對話,並以任何順序展示,而 Gemini 亦可以以自由排序回應。
輸入圖片可以有唔同嘅解像度,而視頻則以一系列圖片輸入。音頻以 16 kHz 取樣,然後通過通用語音模型轉換成一系列嘅 token。Gemini 嘅數據集係多模態同多語言嘅,包括「網頁文檔、書籍同代碼,並包括圖片、音頻同視頻數據」。[42]
Demis Hassabis 聲稱訓練 Gemini 1 用咗「大約同 GPT-4 傳聞所需嘅計算量差唔多,甚至可能稍多」。[43]
第二代嘅 Gemini("Gemini 1.5")目前已發布兩個模型:[44]
- Gemini 1.5 Pro。佢係一個多模態嘅稀疏專家混合模型,上下文長度達到「數百萬」。
- Gemini 1.5 Flash。佢係從 Gemini 1.5 Pro 線上蒸餾[45] 出嚟嘅,上下文長度超過 200 萬。
反響
[編輯]Template:見亦 Gemini 嘅推出喺幾個月嘅激烈揣測同期待之後進行,呢種情況被 MIT Technology Review 形容為「人工智能 hype 嘅巔峰」。[46][47] 喺 2023 年 8 月,研究公司 SemiAnalysis 嘅 Dylan Patel 同 Daniel Nishball 寫咗篇博客文章,宣稱 Gemini 嘅推出會「吞噬世界」,並且超越 GPT-4,促使 OpenAI CEO Sam Altman 喺X(之前叫 Twitter)嘲笑呢對二人。[48][49] 共同創辦 OpenAI 嘅商業大亨 Elon Musk 亦加入討論,問道:「數字錯咗咩?」[50] Business Insider 嘅 Hugh Langley 評論話,Gemini 會係 Google 嘅「生死關頭」,佢寫道:「如果 Gemini 令人驚艷,佢會幫助 Google 改變被 Microsoft 同 OpenAI 陰影所困嘅敘事。如果失望,將會鼓舞批評者話 Google 已經落後。」[51]
對於 2023 年 12 月 Gemini 嘅揭幕,華盛頓大學嘅名譽教授 Oren Etzioni 預測 Google 同 OpenAI 之間會有一場「針鋒相對嘅軍備競賽」。加州大學伯克利分校嘅 Alexei Efros 教授讚揚了 Gemini 嘅多模態方法嘅潛力,[19] 同時 聖塔菲研究所嘅科學家 Melanie Mitchell 稱 Gemini 為「非常複雜」。華盛頓大學嘅 Chirag Shah 教授則較少印象深刻,將 Gemini 嘅推出比擬為Apple 每年推出新 iPhone嘅常規操作。同樣地,斯坦福大學嘅 Percy Liang、華盛頓大學嘅 Emily Bender 同 高威大學嘅 Michael Madden 都警告話,冇了解訓練數據嘅情況下,解讀基準分數係困難嘅。[46][52] 為 Fast Company 撰文嘅 Mark Sullivan 認為 Google 有機會挑戰 iPhone 嘅主導市場份額,佢相信 Apple 可能冇能力開發類似 Gemini 嘅功能,特別係Siri 虛擬助手。[53] Gemini 推出嘅第二日,Google 嘅股價上升咗 5.3% 。[54][55]
Google 因為一段經過編輯嘅 Gemini 示範影片而受到批評,該影片唔係即時錄製嘅。[56]
參見
[編輯]參考資料
[編輯]{{reflist|refs=
[55] ```
- ↑ 1.0 1.1 Grant, Nico (May 10, 2023). "Google Builds on Tech's Latest Craze With Its Own A.I. Products". The New York Times. ISSN 0362-4331. 原先內容歸檔喺May 10, 2023. 喺August 21, 2023搵到.
- ↑ 2.0 2.1 Ortiz, Sabrina (May 10, 2023). "Every major AI feature announced at Google I/O 2023". ZDNet. 原先內容歸檔喺May 10, 2023. 喺August 21, 2023搵到.
- ↑ 3.0 3.1 Milmo, Dan (December 6, 2023). "Google says new AI model Gemini outperforms ChatGPT in most tests". The Guardian. ISSN 0261-3077. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ 4.0 4.1 Levy, Steven (September 11, 2023). "Sundar Pichai on Google;s AI, Microsoft's AI, OpenAI, and ... Did We Mention AI?". Wired. 原先內容歸檔喺September 11, 2023. 喺September 12, 2023搵到.
- ↑ 5.0 5.1 Knight, Will (June 26, 2023). "Google DeepMind's CEO Says Its Next Algorithm Will Eclipse ChatGPT". Wired. 原先內容歸檔喺June 26, 2023. 喺August 21, 2023搵到.
- ↑ 6.0 6.1 6.2 6.3 Victor, Jon (August 15, 2023). "How Google is Planning to Beat OpenAI". The Information. 原先內容歸檔喺August 15, 2023. 喺August 21, 2023搵到.
- ↑ 7.0 7.1 Grant, Nico (January 20, 2023). "Google Calls In Help From Larry Page and Sergey Brin for A.I. Fight". The New York Times. ISSN 0362-4331. 原先內容歸檔喺January 20, 2023. 喺February 6, 2023搵到.
- ↑ 8.0 8.1 Kruppa, Miles; Seetharaman, Deepa (July 21, 2023). "Sergey Brin Is Back in the Trenches at Google". The Wall Street Journal. ISSN 0099-9660. 原先內容歸檔喺July 21, 2023. 喺September 7, 2023搵到.
- ↑ 9.0 9.1 Carter, Tom (December 7, 2023). "Google confirms that its cofounder Sergey Brin played a key role in creating its ChatGPT rival". Business Insider. 原先內容歸檔喺December 7, 2023. 喺December 31, 2023搵到.
- ↑ 10.0 10.1 Victor, Jon (September 18, 2023). "OpenAI Hustles to Beat Google to Launch 'Multimodal' LLM". The Information. 原先內容歸檔喺September 18, 2023. 喺October 15, 2023搵到.
- ↑ 11.0 11.1 "Google nears release of AI software Gemini, The Information reports". Reuters. September 14, 2023. 原先內容歸檔喺September 15, 2023. 喺October 2, 2023搵到.
- ↑ 12.0 12.1 Nolan, Beatrice (September 23, 2023). "Google is quietly handing out early demos of its GPT-4 rival called Gemini. Here's what we know so far about the upcoming AI model". Business Insider. 原先內容歸檔喺September 23, 2023. 喺October 16, 2023搵到.
- ↑ 13.0 13.1 13.2 13.3 Kruppa, Miles (December 6, 2023). "Google Announces AI System Gemini After Turmoil at Rival OpenAI". The Wall Street Journal. ISSN 0099-9660. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ 14.0 14.1 14.2 14.3 14.4 Liedtike, Michael; O'Brien, Matt (December 6, 2023). "Google launches Gemini, upping the stakes in the global AI race". Associated Press. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ 15.0 15.1 Edwards, Benj (December 6, 2023). "Google launches Gemini—a powerful AI model it says can surpass GPT-4". Ars Technica. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ 16.0 16.1 16.2 Pierce, David (December 6, 2023). "Google launches Gemini, the AI model it hopes will take down GPT-4". The Verge. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ 17.0 17.1 Fung, Brian; Thorbecke, Catherine (December 6, 2023). "Google launches Gemini, its most-advanced AI model yet, as it races to compete with ChatGPT". CNN Business. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ 18.0 18.1 "Google launches Gemini, upping the stakes in the global AI race". CBS News. December 6, 2023. 原先內容歸檔喺December 7, 2023. 喺December 7, 2023搵到.
- ↑ 19.0 19.1 19.2 Knight, Will (December 6, 2023). "Google Just Launched Gemini, Its Long-Awaited Answer to ChatGPT". Wired. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ 20.0 20.1 引用錯誤 無效嘅
<ref>
標籤;無文字提供畀叫做AnnounceTime2
嘅參照 - ↑ 21.0 21.1 引用錯誤 無效嘅
<ref>
標籤;無文字提供畀叫做AnnounceWSJ2
嘅參照 - ↑ 22.0 22.1 22.2 引用錯誤 無效嘅
<ref>
標籤;無文字提供畀叫做AnnounceGuardian2
嘅參照 - ↑ 引用錯誤 無效嘅
<ref>
標籤;無文字提供畀叫做AnnounceWired2
嘅參照 - ↑ 引用錯誤 無效嘅
<ref>
標籤;無文字提供畀叫做AnnounceNYT2
嘅參照 - ↑ 引用錯誤 無效嘅
<ref>
標籤;無文字提供畀叫做AnnounceZDNet
嘅參照 - ↑ 引用錯誤 無效嘅
<ref>
標籤;無文字提供畀叫做AnnounceArs2
嘅參照 - ↑ 引用錯誤 無效嘅
<ref>
標籤;無文字提供畀叫做AnnounceAP2
嘅參照 - ↑ 28.0 28.1 Gurman, Mark; Love, Julia; Alba, Davey (January 17, 2024). "Samsung Bets on Google-Powered AI Features in Smartphone Revamp". Bloomberg News. 原先內容歸檔喺January 17, 2024. 喺February 4, 2024搵到.
- ↑ 29.0 29.1 Chokkattu, Julian (January 17, 2024). "Samsung's Galaxy S24 Phones Call on Google's AI to Spruce Up Their Smarts". Wired. 原先內容歸檔喺January 17, 2024. 喺February 4, 2024搵到.
- ↑ 30.0 30.1 Metz, Cade (February 8, 2024). "Google Releases Gemini, an A.I.-Driven Chatbot and Voice Assistant". The New York Times. ISSN 0362-4331. 原先內容歸檔喺February 8, 2024. 喺February 8, 2024搵到.
- ↑ 31.0 31.1 Dastin, Jeffrey (February 8, 2024). "Google rebrands Bard chatbot as Gemini, rolls out paid subscription". Reuters. 原先內容歸檔喺February 8, 2024. 喺February 8, 2024搵到.
- ↑ 32.0 32.1 Li, Abner (February 8, 2024). "Google One AI Premium is $19.99/mo with Gemini Advanced & Gemini for Workspace". 9to5Google. 原先內容歸檔喺February 8, 2024. 喺February 8, 2024搵到.
- ↑ 33.0 33.1 Mehta, Ivan (February 1, 2024). "Google's Bard chatbot gets the Gemini Pro update globally". TechCrunch. 原先內容歸檔喺February 1, 2024. 喺February 4, 2024搵到.
- ↑ 34.0 34.1 Knight, Will (February 15, 2024). "Google's Flagship AI Model Gets a Mighty Fast Upgrade". Wired. 原先內容歸檔喺February 15, 2024. 喺February 21, 2024搵到.
- ↑ 35.0 35.1 Nieva, Richard (February 15, 2024). "Google Unveils Gemini 1.5, But Only Developers And Enterprise Clients Have Access For Now". Forbes. 原先內容歸檔喺February 15, 2024. 喺February 21, 2024搵到.
- ↑ 36.0 36.1 McCracken, Harry (February 15, 2024). "Google's new Gemini 1.5 AI can dive deep into oceans of video and audio". Fast Company. 原先內容歸檔喺February 17, 2024. 喺February 21, 2024搵到.
- ↑ 37.0 37.1 Stokes, Samantha (February 15, 2024). "Here's everything you need to know about Gemini 1.5, Google's newly updated AI model that hopes to challenge OpenAI". Business Insider. 原先內容歸檔喺February 19, 2024. 喺February 21, 2024搵到.
- ↑ 38.0 38.1 Khan, Jeremy (February 21, 2024). "Google unveils new family of open-source AI models called Gemma to take on Meta and others—deciding open-source AI ain't so bad after all". Fast Company. 原先內容歸檔喺February 21, 2024. 喺February 21, 2024搵到.
- ↑ 39.0 39.1 Alba, Davey (February 21, 2024). "Google Delves Deeper Into Open Source with Launch of Gemma AI Model". Bloomberg News. 原先內容歸檔喺February 21, 2024. 喺February 21, 2024搵到.
- ↑ 40.0 40.1 Metz, Cade; Grant, Nico (February 21, 2024). "Google Is Giving Away Some of the A.I. That Powers Chatbots". The New York Times. ISSN 0362-4331. 原先內容歸檔喺February 21, 2024. 喺February 21, 2024搵到.
- ↑ "Gemini Flash". Google DeepMind (英文). May 14, 2024. 原先內容歸檔喺July 11, 2024. 喺2024-05-17搵到.
- ↑ 42.0 42.1 Template:Cite tech report
- ↑ Patel, Dwarkesh (2024-02-21). "Demis Hassabis - Scaling, Superhuman AIs, AlphaZero atop LLMs, Rogue Nations Threat". www.dwarkeshpatel.com (英文). 原先內容歸檔喺March 3, 2024. 喺2024-03-03搵到.
- ↑ 44.0 44.1 Template:Cite tech report
- ↑ Anil, Rohan; Pereyra, Gabriel; Passos, Alexandre; Ormandi, Robert; Dahl, George E.; Hinton, Geoffrey E. (2020-08-20). "Large scale distributed neural network training through online distillation". arXiv:1804.03235 [cs.LG].
- ↑ 46.0 46.1 46.2 Heikkilä, Melissa; Heaven, Will Douglas (December 6, 2023). "Google DeepMind's new Gemini model looks amazing—but could signal peak AI hype". MIT Technology Review. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ 47.0 47.1 Henshall, Will (December 6, 2023). "Google DeepMind Unveils Its Most Powerful AI Offering Yet". Time. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ 48.0 48.1 howdhury, Hasan (August 29, 2023). "AI bros are at war over declarations that Google's upcoming Gemini AI model smashes OpenAI's GPT-4". Business Insider. 原先內容歸檔喺August 29, 2023. 喺September 7, 2023搵到.
- ↑ 49.0 49.1 Harrison, Maggie (August 31, 2023). "OpenAI Rages at Report that Google's New AI Crushes GPT-4". Fortune. 原先內容歸檔喺August 31, 2023. 喺September 7, 2023搵到.
- ↑ 50.0 50.1 Musk, Elon [@elonmusk] (August 29, 2023). "Are the numbers wrong?" (推文). 原先內容歸檔喺September 1, 2023. 喺October 15, 2023搵到 –透過Twitter.
- ↑ 51.0 51.1 Langley, Hugh (October 12, 2023). "Google VP teases Gemini's multimodal future: 'I've seen some pretty amazing things.'". Business Insider. 原先內容歸檔喺October 12, 2023. 喺October 15, 2023搵到.
- ↑ 52.0 52.1 Madden, Michael G. (December 15, 2023). "Google's Gemini: is the new AI model really better than ChatGPT?". The Conversation. 原先內容歸檔喺December 15, 2023. 喺February 4, 2024搵到.
- ↑ 53.0 53.1 Sullivan, Mark (December 6, 2023). "Gemini-powered Google phones may make Siri even more of an Achilles' heel for the iPhone". Fast Company. 原先內容歸檔喺December 7, 2023. 喺December 7, 2023搵到.
- ↑ 54.0 54.1 Soni, Aditya (December 7, 2023). "Alphabet soars as Wall Street cheers arrival of AI model Gemini". Reuters. 原先內容歸檔喺December 7, 2023. 喺February 4, 2024搵到.
- ↑ 55.0 55.1 Swartz, Jon (December 7, 2023). "Gemini, Google's long-awaited answer to ChatGPT, is an overnight hit". MarketWatch. 原先內容歸檔喺December 7, 2023. 喺February 4, 2024搵到.
- ↑ Elias, Steve Kovach, Jennifer (2023-12-08). "Google faces controversy over edited Gemini AI demo video". CNBC (英文). 原先內容歸檔喺December 9, 2023. 喺2023-12-09搵到.
- ↑ Elias, Jennifer (December 6, 2023). "Google launches its largest and 'most capable' AI model, Gemini". CNBC. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ Metz, Cade; Grant, Nico (December 6, 2023). "Google Updates Bard Chatbot With 'Gemini' A.I. as It Chases ChatGPT". The New York Times. ISSN 0362-4331. 原先內容歸檔喺December 6, 2023. 喺December 6, 2023搵到.
- ↑ Alba, Davey; Ghaffary, Shirin (December 6, 2023). "Google Opens Access to Gemini, Racing to Catch Up to OpenAI". Bloomberg News. 原先內容歸檔喺December 6, 2023. 喺December 7, 2023搵到.
- ↑ Knight, Will (December 6, 2023). "Google DeepMind's Demis Hassabis Says Gemini Is a New Breed of AI". Wired. 原先內容歸檔喺December 6, 2023. 喺December 7, 2023搵到.