谷歌在周三凌晨的Made by Google活动上，正式发布了其最新的AI语音助手Gemini Live，以及一系列搭载Google Tensor G4芯片的Pixel硬件产品。在OpenAI的“草莓大模型”即GPT-4o尚未进入iPhone之前，谷歌抢先一步完成了手机版的落地。Gemini Live对标的是OpenAI的高级语音模式，提供了与ChatGPT相似的移动对话体验，支持用户与AI助手进行自由流畅的对话，甚至可以像在普通电话中一样打断或改变话题，无需打字。
Gemini Live还提供了10种新的自然声音供用户选择，而OpenAI仅提供3种声音。用户可以按照自己的节奏说话，或者在回答过程中打断AI并提出其他问题，就像在平时对话中一样。Gemini Live是直接唤醒的，可以在应用程序在后台运行或手机锁屏时继续与AI对话，且对话可以随时暂停和恢复。此外，Gemini Live还将与多种安卓应用的功能集成，提高AI的可用性。
此次发布的硬件产品包括Pixel 9、Pixel 9 Pro和Pixel 9 Pro XL，以及一款折叠屏手机Pixel 9 Pro Fold。这些手机均由全新的Google Tensor G4芯片提供支持，可以带来各种生成式AI能力。Pixel 9手机采用全新外观，摄像头置于正面和中心位置，改进了标志性的摄像头模组，提升了手感。谷歌宣称，这些手机的耐用性是Pixel 8的两倍。
虽然活动现场演示中遇到了一些问题，比如两次让手机识图失败，但最终Gemini Live成功从图片中提取相关信息并连接日历，为用户提供了准确结果。谷歌产品经理Leland Rechis介绍，谷歌不允许Gemini Live模仿这10种声音以外的任何声音，以避免与版权法发生冲突。
总体而言，Gemini Live似乎是一种比使用简单的Google搜索更自然地深入研究主题的好方法。谷歌表示，Gemini Live是Project Astra迈出的一步，Project Astra是该公司在Google I/O期间首次亮相的多模态AI模型。未来，谷歌希望增加实时视频理解功能。
随着Gemini Live的推出和谷歌硬件全家桶的到来，谷歌在AI手机领域又迈出了重要一步，与苹果和华为等竞争对手展开了更激烈的竞争。
英语如下：
News Title: “Google’s New AI Phone Stuns Market with GPT-4o Integration”
Keywords: Google AI Phone, Gemini Live, GPT-4o Launch
News Content:
Google unveiled its latest AI voice assistant, Gemini Live, and a series of Pixel hardware products powered by the Google Tensor G4 chip at the Made by Google event on Wednesday morning. Before GPT-4o from OpenAI could be integrated into the iPhone, Google took the lead in bringing the technology to mobile devices. Gemini Live competes with OpenAI’s advanced voice mode, offering a mobile conversational experience similar to ChatGPT, allowing users to engage in free-flowing, voice-based interactions with AI assistants, even interrupting or changing topics as one would in a regular phone call, without the need for typing.
Gemini Live offers 10 new natural-sounding voices for users to choose from, compared to OpenAI’s three options. Users can speak at their own pace or interrupt the AI to ask follow-up questions, just as in everyday conversation. Gemini Live is directly wakeable, allowing users to continue conversations with the AI while apps are running in the background or when the phone is locked, and the conversation can be paused and resumed at any time. Additionally, Gemini Live will integrate with various Android applications to enhance the AI’s usability.
The hardware products released include the Pixel 9, Pixel 9 Pro, and Pixel 9 Pro XL, as well as a foldable phone, the Pixel 9 Pro Fold. All these devices are supported by the new Google Tensor G4 chip, bringing various generative AI capabilities. The Pixel 9 features a new design with the camera positioned at the front and center, improving the iconic camera module and enhancing the tactile experience. Google claims that these phones are twice as durable as the Pixel 8.
Although there were some issues during the live demonstration, such as two failed attempts to recognize images, Gemini Live ultimately succeeded in extracting relevant information from the images and connecting to the user’s calendar, providing accurate results. Google product manager Leland Rechis explained that Google does not allow Gemini Live to mimic any voices outside of these 10, to avoid conflicts with copyright laws.
Overall, Gemini Live appears to be a more natural way to delve into topics than using simple Google search. Google stated that Gemini Live is a step forward for Project Astra, a multimodal AI model first introduced at Google I/O. In the future, Google aims to add real-time video understanding capabilities.
With the launch of Gemini Live and Google’s hardware lineup, the company has taken a significant step in the AI phone market, intensifying competition with rivals like Apple and Huawei.
【来源】https://www.jiqizhixin.com/articles/2024-08-14-7
Views: 2