Google 全新「任意對任意」AI 模型 Gemini Omni 實測：效果驚人且近乎無縫
Original: Google’s new anything-to-anything AI model is wild
Google recently unveiled a brand-new "anything-to-anything" multimodal AI model — Gemini Omni — whose powerful cross-modal generation and…
Google 發表了全新的「任意對任意（anything-to-anything）」AI 模型 Gemini Omni。外媒記者實測將其用於將小孩的毛絨鹿玩具「Buddy」合成到各種度假場景中，發現其生成效果極其逼真且操作簡單。這款模型不僅展現了強大的多模態影片生成與編輯能力，同時也再度引發了關於深偽（Deepfake）技術門檻降低與倫理界線的討論。
Google recently unveiled a brand-new "anything-to-anything" multimodal AI model — Gemini Omni — whose powerful cross-modal generation and transformation capabilities have drawn intense attention from the tech world. A journalist from an overseas outlet shared their hands-on experience with the model. The author mentioned that last year they had tried to recreate a scene from a Google advertisement by using deepfake technology to produce a video of their four-year-old son's stuffed deer toy, "Buddy," appearing to vacation in various locations around the world — but at the time, the technical barriers and output quality were still quite limited.
Free shows the 3-line summary; Pro unlocks the full deep summary (~300 words) so you never have to click through.
See Pro plans →Want the original English / full article?
Read on The Verge AI →Summaries are AI-generated; the original article is authoritative.