Tavus 推出实时视频对话模型 Griffin，48% 测试者误认其为真人
Nearly half of test subjects mistook Tavus' AI video avatar for a real person on a one-minute call
Tavus 推出名为 Griffin 的 Human Interaction Model，可在实时视频通话中同时接收和生成语音、面部表情、语气、手势与停顿。Tavus 研究显示，48% 的参与者在一次一分钟视频通话后认为 Griffin 是真人，此前系统最高为 2%。在 Tavus 称为 Nvidia 独立测试的拟人度评分中，Griffin 得 3.83 分，真人为 3.92 分，此前最佳 AI 模型为 2.80 分。名为 Griffin-Lite 的预览版已向部分测试者开放，能力更强的版本将在安全问题解决后推出。
正文
当前显示原文，中文全文译文尚未就绪。
Tavus has introduced Griffin, what the company calls the first "Human Interaction Model" (HIM), a class of model designed to understand and carry on face-to-face conversations in real time. Griffin processes speech, facial expressions, tone of voice, gestures, and pauses while both receiving and generating video.
In a Tavus study, 48 percent of participants believed Griffin was a real person after a one-minute video call. Previous systems maxed out at two percent. In what Tavus describes as an independent Nvidia test measuring how human an AI feels in direct audio-video conversation, Griffin scored 3.83 points. Actual humans scored 3.92, the previous best AI model hit 2.80.
A preview called Griffin-Lite is available to "select testers as a research preview," with a more capable version coming once "safety concerns are addressed." Tavus lists tutoring, practicing difficult conversations, and camera-based tech support as potential use cases.
The San Francisco AI startup was founded in 2020 and has raised about $64 million. It started out building personalized AI videos for sales and marketing, then expanded into live video conversations with digital personas . More details on Griffin are in the Tavus research report .